School of Chemical Technology Degree Programme of Chemical Technology
Laura Suominen
IMPURITY REMOVAL OF VACUUM GAS OIL
Master’s thesis for the degree of Master of Science in Technology
submitted for inspection, Espoo, 22 May, 2016.
Supervisor Professor Pekka Oinas
Instructors D. Sc. (Tech.) Aarne Sundberg
D. Sc. (Tech.) Sarwar Golam
Aalto University, P.O. BOX 11000, 00076 AALTO
www.aalto.fi
Abstract of master's thesis
Author Laura Suominen
Title of thesis Impurity removal of vacuum gas oil
Department Department of Biotechnology and Chemical Technology
Professorship Plant Design Code of professorship KE-107
Thesis supervisor Professor Pekka Oinas
Thesis advisors/Thesis examiners D. Sc. (Tech.) Aarne Sundberg, D. Sc. (Tech.)
Sarwar Golam
Date 22.5.2016 Number of pages 107+1 Language English
Abstract
Generally, vacuum gas oil (VGO) is defined as a petroleum distillate boiling between 350 and 550 °C at 1 atm. VGO is produced from atmospheric residue in a vacuum distillation unit. The amount and nature of impurities in VGO depend greatly on the origin of the crude oil but the distillation range of the petroleum cut also affects the impurity level. Typical impurities are nitrogen, sulphur and metals such as nickel, vanadium and iron. As crude oil typically has a large excess of heavy cuts and an insufficient amount of light cuts and middle distillates, the heaviest cuts need to be converted into lighter products. Before VGO is converted into more valuable lighter products, it is fed to a catalytic hydrodesulphurization (HDS) process to reduce the sulphur levels. However, the impurities present in VGO lead to deactivation of HDS catalyst. Therefore, an effective impurity removal method prior to HDS process would extend the catalyst lifetime and reduce refinery downtime. The impurity removal methods discussed in this thesis are solvent deasphalting, demetallization by acids, phosphorous compounds and supercritical water, and photochemical denitrogenation and desulphurization. Based on the advantages and disadvantages of the methods, demetallization using phosphorous compounds was chosen to be preliminarily designed in the applied part of the thesis. The VGO feed contained 10 ppm of metal impurities. When no demetallization pretreatment was applied, the HDS catalyst lifetime was only 8 weeks. In the demetallization process phosphoric acid was used as the phosphorous compound. The acid was mixed with the VGO in a static mixer and the phases were separated in three parallel decanters. Phosphoric acid reduced the concentration of Ni, V and Fe and the removal rates used in the calculations were 32 w%, 23 w% and 22 w%, respectively. As a result, the HDS catalyst lifetime was 9 weeks. The catalyst lifetime was extended only by one week and the demetallization process was not profitable with current costs.
Keywords vacuum gas oil, VGO, impurity removal, hydrodesulphurization, HDS,
demetallization, phosphoric acid
Aalto-yliopisto, PL 11000, 00076 AALTO
www.aalto.fi
Diplomityön tiivistelmä
Tekijä Laura Suominen
Työn nimi Tyhjökaasuöljyn epäpuhtauksien poisto
Laitos Biotekniikan ja kemian tekniikan laitos
Professuuri Tehdassuunnittelu Professuurikoodi KE-107
Työn valvoja Professori Pekka Oinas
Työn ohjaajat/Työn tarkastajat TkT Aarne Sundberg, TkT Sarwar Golam
Päivämäärä 22.5.2016 Sivumäärä 107+1 Kieli Englanti
Tiivistelmä
Yleisesti tyhjökaasuöljy (vacuum gas oil, VGO) määritellään raakaöljyjakeeksi, joka tislautuu ilmanpaineessa 350–550 °C:n lämpötilassa. Se valmistetaan tyhjötislaamalla pohjaöljystä. VGO:n sisältämien epäpuhtauksien määrä ja laatu riippuu suuresti raakaöljyn alkuperästä, mutta myös jakeen tislausväli vaikuttaa epäpuhtaustasoon. Tyypillisiä epäpuhtauksia ovat typpi, rikki ja metallit, kuten nikkeli, vanadiini ja rauta. Koska raakaöljy tyypillisesti sisältää liikaa raskaita jakeita ja riittämättömästi kevyitä jakeita ja keskitisleitä, raskaimmat jakeet jalostetaan kevyemmiksi tuotteiksi. Ennen tätä VGO:n sisältämä rikki poistetaan katalyyttisesti vedyn avulla (hydrodesulphurization, HDS). VGO:n sisältämät epäpuhtaudet kuitenkin deaktivoivat rikinpoistossa käytetyn katalyytin. Tästä johtuen tehokas epäpuhtauksien poistomenetelmä ennen rikinpoistoprosessia pidentäisi katalyytin elinikää ja vähentäisi seisokkien tarvetta. Tässä työssä käsiteltyjä epäpuhtauksien poistomenetelmiä ovat deasfaltointi, metallinpoisto happojen, fosforiyhdisteiden ja ylikriittisen veden avulla sekä fotokemiallinen typen- ja rikinpoisto. Menetelmien etujen ja haittojen perusteella esisuunnitteluun valittiin metallinpoisto fosforiyhdisteiden avulla. Prosessin esisuunnittelu toteutettiin työn soveltavassa osassa. VGO-syöttö sisälsi 10 ppm metalliepäpuhtauksia. Kun metallinpoistoa ei ollut, rikinpoistokatalyytin elinikä oli ainoastaan 8 viikkoa. Metallinpoistoprosessissa fosforiyhdisteenä käytettiin fosforihappoa. Happo sekoitettiin VGO-virtaan staattisessa sekoittimessa ja faasit erotettiin toisistaan kolmessa rinnakkaisessa dekantterissa. Fosforihappo vähensi nikkelin, vanadiinin ja raudan pitoisuutta ja laskuissa käytetyt poistoprosentit olivat nikkelille 32 m-%, vanadiinille 23 m-% ja raudalle 22 m-%. Tämän seurauksena rikinpoistokatalyytin elinikä oli 9 viikkoa. Katalyytin elinikä piteni vain yhdellä viikolla eikä metallinpoistoprosessi ollut kannattava nykyisillä kustannuksilla.
Avainsanat tyhjökaasuöljy, VGO, epäpuhtauksien poisto, rikinpoisto vedyllä, HDS,
metallinpoisto, fosforihappo
Preface
This master’s thesis was conducted for Neste in the Technology Centre of Porvoo refinery
between October 2015 and March 2016. The thesis was supervised by professor Pekka
Oinas from Aalto University.
I would like to thank my advisors D. Sc. (Tech.) Aarne Sundberg and D. Sc. (Tech.) Sarwar
Golam for the guidance they gave me during my work. I would also like to thank Jaana
Makkonen who was responsible of the project for which my thesis was done. She offered
help and encouraging comments throughout the process. Furthermore, I wish to thank
Prof. Pekka Oinas for reviewing my thesis and giving me valuable advice.
Finally, I want to thank my family and friends. I am thankful for the support and
encouragement of my family and for all the great moments with my friends that made
my time in Otaniemi so memorable.
Porvoo, March 24th, 2016
Laura Suominen
Table of Contents
1 Introduction .................................................................................................................... 1
1.1 Goals of the thesis ................................................................................................... 1
1.2 Scope of the thesis ................................................................................................... 2
LITERATURE PART .............................................................................................................. 3
2 Vacuum gas oil ................................................................................................................ 3
3 Impurities in VGO ............................................................................................................ 6
3.1 Nitrogen ................................................................................................................... 7
3.2 Sulphur ..................................................................................................................... 8
3.3 Asphaltenes .............................................................................................................. 9
3.4 Nickel and vanadium .............................................................................................. 10
3.5 Carbon residue ....................................................................................................... 12
3.6 Iron ......................................................................................................................... 13
3.7 Calcium ................................................................................................................... 14
3.8 Silicon ..................................................................................................................... 15
4 Impurity removal methods ........................................................................................... 16
4.1 Physical method ..................................................................................................... 16
4.1.1 Solvent deasphalting ....................................................................................... 16
4.2 Chemical methods ................................................................................................. 20
4.2.1 Demetallization by acids ................................................................................. 20
4.2.2 Demetallization by phosphorous compounds ................................................ 25
4.2.3 Demetallization by supercritical water ........................................................... 28
4.2.4 Photochemical denitrogenation and desulphurization .................................. 32
5 Advantages and disadvantages of the methods ........................................................... 36
5.1 Comparison of the methods .................................................................................. 36
5.2 Selection of one process for prestudy ................................................................... 42
APPLIED PART .................................................................................................................. 43
6 Designing of an impurity removal unit ......................................................................... 43
7 Case 1. HDS process without a new impurity removal system .................................... 46
7.1 Process description ................................................................................................ 46
7.2 Catalyst lifetime ..................................................................................................... 47
8 Case 2. Impurity removal integrated in the HDS process ............................................. 49
8.1 Process design ........................................................................................................ 49
8.1.1 Process flow chart ........................................................................................... 49
8.1.2 Mass flow rate of phosphoric acid.................................................................. 51
8.1.3 Physical properties of VGO and phosphoric acid ........................................... 53
8.2 Equipment .............................................................................................................. 54
8.2.1 Static mixer ..................................................................................................... 54
Dimensions of the mixer ...................................................................................... 55
Pressure drop ....................................................................................................... 58
Materials .............................................................................................................. 61
Wall thickness ...................................................................................................... 62
8.2.2 Decanter .......................................................................................................... 64
Dimensions of the decanter................................................................................. 65
Materials .............................................................................................................. 70
Wall thickness ...................................................................................................... 71
Decanter heads .................................................................................................... 73
Decanter support ................................................................................................. 76
8.2.3 Electrostatic coalescer .................................................................................... 77
8.2.4 Storage tank for phosphoric acid .................................................................... 80
8.3 Profitability ............................................................................................................ 81
8.3.1 Investment cost .............................................................................................. 81
Static mixer .......................................................................................................... 81
Decanter ............................................................................................................... 83
8.3.2 Operating costs ............................................................................................... 87
Purchase cost of VGO and phosphoric acid ......................................................... 87
Cost of changing the catalyst ............................................................................... 87
8.3.3 Sensitivity analysis .......................................................................................... 89
8.4 Safety ..................................................................................................................... 93
9 Discussion...................................................................................................................... 96
10 Conclusions ................................................................................................................. 97
Bibliography ..................................................................................................................... 99
APPENDICES
Appendix 1. Graph for determining the friction factor in pipes.
Abbreviations
ASME American Society of Mechanical Engineers
BPVC Boiler & Pressure Vessel Code
CEPCI Chemical Engineering Plant Cost Index
DAO Deasphalted oil
HDS Hydrodesulphurization
LHSV Liquid Hourly Space Velocity
NFPA National Fire Protection Association
NPV Net present value
PDMS Polydimethylsiloxane
SCW Supercritical water
SDA Solvent deasphalting
TAN Total acid number
VGO Vacuum gas oil
1
1 Introduction
Importance of upgrading heavy oils is increasing as heavier crudes are more often utilized
to meet the demand of lighter oils. Heavy oils contain significant quantities of metal,
sulphur and nitrogen contaminants which complicate the upgrading process. The metals,
mainly vanadium and nickel, pose a particular problem for refineries as they accumulate
on catalysts causing permanent deactivation. (Mandal et al., 2012)
Sulphur contaminants present in petroleum can be removed in a catalytic
hydrodesulphurization (HDS) process. Although the process is capable of selectively
removing the contaminants, a major drawback is the formation of metal deposits on the
catalyst. Therefore, a guard bed consisting of a cheaper hydrogenation catalyst is used
to prevent the deactivation of desulphurization catalyst. Due to the fast deactivation of
the guard bed catalyst, it needs to be changed regularly to maintain the activity. (Fahim
et al., 2010)
Alternative impurity removal methods have been used and studied (Ali & Abbas, 2006).
By removing the contaminants in a pretreatment process before desulphurization, the
formation of deposits on the catalyst surface would decrease. As a result, the catalyst
deactivation would be reduced and the heavier petroleum fractions with high impurity
levels could be more cost-efficiently upgraded.
1.1 Goals of the thesis
The aim of this thesis was to study possible impurity removal methods for vacuum gas
oil prior to the hydrodesulphurization process. An impurity removal pretreatment could
enable the use of cheaper, poor-quality vacuum gas oil which has high impurity levels.
By decreasing the impurity levels, the catalyst lifetime could be extended as without any
pretreatment, the catalyst would be rapidly deactivated by the high amount of
2
impurities. As a result, the catalyst could be replaced less frequently reducing refinery
downtime and leading to significant financial savings.
1.2 Scope of the thesis
In the literature part of this thesis, the most common impurities present in the vacuum
gas oil are presented and their effects are discussed. Different impurity removal methods
were studied excluding adsorbent materials and guard beds. Comparison of the methods
was conducted and based on their advantages and disadvantages, one of the impurity
removal processes was chosen for prestudy. The prestudy was carried out in the applied
part of the thesis and includes preliminary design and sizing of process equipment, cost
estimation and discussion about safety. In addition, the HDS process with the impurity
removal was compared to the HDS process without any pretreatment and a sensitivity
analysis was conducted.
3
LITERATURE PART
2 Vacuum gas oil
Vacuum gas oil (VGO) is a petroleum distillate boiling between 350 and 550 °C at 1 atm
(Leprince, 2001; Toulhoat & Raybaud, 2013). It is produced from atmospheric residue in
a vacuum distillation unit. VGO is further processed to produce light and middle
distillates. (Parkash, 2003) A flow scheme of crude oil refining is presented in Figure 1
where vacuum distillation can be seen in relation to other refinery units.
Figure 1. Oil refinery flow scheme (Fahim et al., 2010).
4
Crude oil entering a refinery is first distilled in an atmospheric distillation tower in order
to fractionate the hydrocarbons. Due to the fragility of some molecules which start to
crack at around 380-400 °C, the atmospheric distillation can fractionate hydrocarbons
with boiling point below 350-380 °C. Distilling heavier fractions from crude oil is
impossible at atmospheric pressure. Therefore, distillation is continued under vacuum
to stay within the temperature range that the compounds can withstand. (Marcilly,
2006)
The heaviest cut distilling above 350-380 °C forms the atmospheric residue which is used
as the feed to the vacuum distillation. In a vacuum distillation tower, the atmospheric
residue is fractionated into vacuum gas oil and vacuum residue. (Marcilly, 2006; Fahim
et al., 2010) The vacuum is produced by steam ejectors taking suction from the top of
the distillation tower. The pressure at the top of the tower is 0.4-0.7 kPa and in the flash
zone about 3-4 kPa. (Jones & Pujadó, 2006) Boiling ranges of VGO and vacuum residue
in relation to other petroleum cuts are shown in Figure 2.
5
Figure 2. Different petroleum cuts with their boiling range and carbon number (Toulhoat & Raybaud, 2013).
However, these distillation cuts of crude oil do not correspond to the market demand of
petroleum products. Crude oil has typically a large excess of heavy cuts and an
insufficient amount of light cuts and middle distillates. To adjust these quantities, the
heaviest cuts need to be converted into lighter products which are in high demand. To
improve the product quality to meet the applicable specifications, some cuts have to
undergo purification treatment and/or restructuring of the molecules. (Marcilly, 2006)
Most crude oils contain traces of different metal complexes. These impurities
concentrate in the residue and therefore, VGO contains several impurities. Nickel and
vanadium are the most abundant and troublesome metal complexes present in the
6
organic parts of fossil fuels deposits. Even though present only in trace quantities, they
lead to deactivation of both desulphurization and cracking catalysts. (Dechaine & Gray,
2010) To prevent the deactivation of the catalysts, various methods for pretreating VGO
have been studied and are presented in chapter 4.
3 Impurities in VGO
The amount and nature of impurities in VGO depend greatly on the origin of the crude
oil. The distillation range of the petroleum cut also affects the impurity level. Typical
impurities are nitrogen, sulphur and metals such as nickel, vanadium and iron. (Leprince,
2001) Typical VGO impurity composition is presented in Table 1.
Table 1. Typical VGO impurity composition (Toulhoat & Raybaud, 2013).
Impurity Unit VGO
Sulphur wt % 0.1–2
Nitrogen wt ppm 300–2000
Basic nitrogen wt ppm 440
Nickel wt ppm < 0.5
Vanadium wt ppm < 0.5
Asphaltenes wt ppm < 100
In this chapter, different impurities and their harmful effects are discussed. The
impurities presented are nitrogen, sulphur, asphaltenes, vanadium, nickel, carbon
residue, iron, calcium and silicon.
7
3.1 Nitrogen
Vacuum gas oil contains different types of nitrogen compounds such as anilines,
compounds containing a five-atom ring (indoles and carbazoles) and compounds
containing a six-atom ring (quinolines and acridines). Nitrogen is predominantly present
in these heterocyclic N-rings. The five- and six-membered N-rings are classified as non-
basic and basic nitrogen compounds, respectively. The non-basic N-rings are based on a
pyrrole core and the basic ones on a pyridine core. (Macaud et al., 2004; Furimsky &
Massoth, 2005) The pyridines are less reactive than pyrroles but they have a stronger
tendency to adsorb on the catalyst (Leprince, 2001). The structures of different nitrogen
compounds are presented in Figure 3.
Figure 3. Nitrogen compounds in petroleum: 1 indole, 2 carbazole, 3 quinoline, 4 acridine and 5 aniline (Macaud et al., 2004).
For hydroprocessing, nitrogen compounds are the most common catalyst poisons due to
their strong adsorption on catalyst sites. They adsorb on catalyst acidic sites: Lewis and
Brønsted sites. Adsorbing on Lewis sites occurs via the N-electron pair or via the
aromatic π-system and on Brønsted sites by interacting with protons. The adsorption can
be reversible or irreversible, depending on reaction conditions. (Furimsky & Massoth,
1999)
8
The basic nitrogen compounds act as strong inhibitors towards the
hydrodesulphurization (HDS) reaction as they poison the acidic sites of the catalysts.
Besides poisoning the hydroprocessing catalysts, the nitrogen compounds are precursors
of coke formation. Sau et al. (2005) studied the inhibiting effect of nitrogen compounds
in crude oil on hydrotreating and hydrocracking reactions. Their results showed that the
inhibitive effects of nitrogen on hydrocracking conversion are highly non-linear and the
conversion rapidly drops as the nitrogen level increases. At higher reaction temperature,
the drop in activity or conversion is less than in lower temperatures. This is a result of
the higher rate of desorption of nitrogen compounds at elevated reaction temperatures.
Selective removal of nitrogen would both increase significantly the activity of
hydrodesulphurization catalyst and reduce the hydrogen consumption. (Sau et al., 2005)
As the 5-membered N-rings are less resistant to hydrogenation compared to the 6-
membered rings, the relative contribution of 5-membered rings to poisoning was
expected to be less important. Therefore, many authors have used pyridine as a poison
in HDS studies. (Furimsky & Massoth, 1999) However, Laredo et al. (2001) studied the
inhibition effect of non-basic nitrogen compounds on HDS of dibenzothiophene. They
discovered that the inhibitive effect of indole and carbazole was comparable to that of
quinoline. The inhibiting effect of these three compounds was very strong even with as
low nitrogen concentrations as 5 ppm. A mixture of indole, carbazole, and quinoline was
found to have a stronger inhibiting effect than any of the individual components at
equivalent concentrations. (Laredo et al., 2001)
3.2 Sulphur
The amount of sulphur present is strongly correlated with the density of the oil cuts.
Therefore, the most of the sulphur compounds are in the heavier cuts. Sulphur is present
as thiols, sulphides, disulphides, sulphoxides, and thiophenes. In heavy fractions, main
sulphur compounds are thiophenes, followed by cyclic and acyclic sulphide derivatives.
9
Sulphoxides are present only in small amounts. (Bertoncini et al., 2013) The structures
of common sulphur compounds are shown in Figure 4.
Figure 4. Structures of sulphur compounds present in VGO: 1. thiol, 2. sulphide, 3. disulphide, and 4. thiophene (Bertoncini et al., 2013).
Sulphur compounds are poisons for noble metal catalysts and in fuels lead to
atmospheric pollutants such as SO2 (Bertoncini et al., 2013). Sulphur oxide gases react
with water in the atmosphere causing acid rain which damages buildings and acidifies
soil. Sulphur emissions also cause respiratory illnesses and contribute to formation of
atmospheric particulates. (Srivastava, 2012) In addition, the presence of sulphur is
undesirable for the reasons of corrosion, bad odor and poor burning. Products containing
high amount of sulphur have low heating values and are considered of poor quality.
(Riazi, 2005)
Environmental regulations are imposing strict limits for sulphur levels in transportation
fuels. Currently, the maximum sulphur content allowed in petrol and diesel in the
European Union is 10 mg/kg. (European Parliament & Council of the European Union,
2009)
3.3 Asphaltenes
Asphaltenes are generally described as n-heptane insoluble components (Gawel et al.,
2005). They are the most complex and biggest molecules in petroleum (Sun et al., 2010).
Asphaltenes consist of condensed polynuclear aromatic layers linked by saturated
10
chains. Their molecular weights vary from a few hundred to several million and they
contain various heteroatoms, such as sulphur, nitrogen, and metals. (Fahim et al., 2010)
Asphaltenes cause problems in the processing of VGO. During hydroprocessing, they
undergo various reactions involving both cracking and hydrogenation which change their
structure. Asphaltenes are the predominant cause of deposit formation. They function
as coke precursors that lead to catalyst deactivation in hydroprocessing. In addition, the
metals present in heavy feedstocks are mainly located in asphaltenes. Therefore, catalyst
deactivation results from the accumulation of both carbonaceous and metallic
deposition. Asphaltenes have a high tendency to adsorb onto the catalyst surface
blocking the pores and preventing other molecules from accessing the active sites.
(Gawel et al., 2005) The effects of metals are discussed in more detail later in this
chapter.
The asphaltene quality is more important than the quantity in coke deposition. For
example, the increase in aromaticity causes higher coke yield. Also increasing the
content of polar components increases the amount of coke deposition. (Gawel et al.,
2005)
3.4 Nickel and vanadium
The most abundant metals are nickel and vanadium. Nickel and vanadium are present in
the form of oil-soluble organometallic complexes. The most common are of porphyrin
type with a nickel or vanadium atom bonded to nitrogen atoms with four heterocyclic
structures. (Speight, 1999; Leprince, 2001) In petroleum, nickel occurs in the +2 valence
state while most vanadium atoms have a valence of +4, almost exclusively as vanadium
ions VO2+. Examples of a nickel and a vanadyl porphyrin complex are presented in Figure
5. Metals are also present in non-porphyrin structures which are not as extensively
studied as the porphyrin structures. (Mandal et al., 2014)
11
Figure 5. Structures of two porphyrins: 1. benzoporphyrin and 2. vanadyl octaethylporphyrin (Speight, 1999).
As the metals accumulate on the catalyst surface, they start to plug up the catalyst pores
leading to catalyst deactivation. The accumulation of metals reduces the access to the
active sites dispersed in the pores of the catalyst. Metals are deposited in the form of
metal sulphides such as Ni3S2 and VS2. They will mainly be deposited in the outer shell of
the catalyst particles so the pore plugging occurs even at relatively low levels of metal
deposition. (Toulhoat & Raybaud, 2013) The metal sulphides gradually narrow the pores
and in order to maintain the design activity, temperature is raised to offset the
deactivation of the catalyst. However, at some point either the temperature is too high
for the reactor design or the loss of active sites by pore plugging is too large to maintain
the design activity by raising the temperature. (Furimsky & Massoth, 1999)
The deactivation by metals is irreversible and it always occurs simultaneously along with
the deactivation by coke. The initial deposition occurs at higher rate for vanadium than
for nickel and increases with larger pore diameter. Therefore, the formation of vanadium
deposits may have an adverse effect on the rate of nickel deposit formation. (Furimsky
& Massoth, 1999) In addition, both nickel and vanadium exhibit dehydrogenation
12
activity. Their presence on the catalyst promotes dehydrogenation reactions during a
cracking process which results in increased amounts of coke and light gases at the
expense of gasoline production. (Ali & Abbas, 2006)
3.5 Carbon residue
Carbon residue is formed by evaporation and thermal degradation of a carbon-
containing material (Rand, 2010). Carbon residue consists of the nonvolatile
carbonaceous compounds of a petroleum fraction when vaporized in the absence of air
at atmospheric pressure. The heavier fractions with more aromatics have higher carbon
residues than light and volatile fractions. (Riazi, 2005)
Higher carbon residue values indicate low-quality fuel and lower hydrogen content.
Moreover, carbon residue increases when sulphur, nitrogen or asphaltenes content or
the viscosity of the oil increases. When hydrogen content increases, the carbon residue
decreases. (Riazi, 2005)
The residue is not composed entirely of carbon. It is coke that can be further processed
by carbon pyrolysis. (Rand, 2010) Coke is one of the main reasons for the deactivation of
hydrotreatment catalysts. Particularly, it leads to plugging which can cause significant
losses of the initial surface area. Coke consists of a deposit of hydrocarbons which
agglomerate to form a film of varying thickness on the surface of a catalyst. (Toulhoat &
Raybaud, 2013)
In petroleum hydrotreatment processes, catalyst deactivation by coke and metal
deposits occurs simultaneously. As can be seen in Figure 6, coke deposits rapidly during
the initial stages before reaching a steady-state. Concurrently, metals exhibit relatively
linear deposition patterns with time. (Furimsky & Massoth, 1999; Toulhoat & Raybaud,
2013)
13
Figure 6. Coke and metal concentrations and the catalyst activity evolution in residue hydrotreatment (Toulhoat & Raybaud, 2013).
The coke build-up increases with molecular weight and boiling range of the feed. The
main reactions leading to coke formation are polymerization and polycondensation.
Among the hydrocarbon groups, alkenes, aromatics and heterocyclics are the most
susceptible to coke formation. They are more likely to convert to higher molecular
weight compounds than the saturated hydrocarbons. (Furimsky & Massoth, 1999)
3.6 Iron
Iron compounds present in VGO originate not only from the crude oil but also from
corrosion. Similarly to nickel and vanadium, iron is present in the structures of porphyrin
compounds. In addition to that, iron is complexed as iron naphthenate. Under
hydrotreating conditions, these complexes are decomposed on the catalyst surface due
to the reaction with hydrogen. Thus, iron is deposited in the form of iron sulphide in the
14
catalyst pores. (Leprince, 2001) The results of iron deposits are the blocking of active
sites and loss of porosity (Furimsky & Massoth, 1999).
Present in crude oils, naphthenic acids are saturated cyclic compounds containing one
or more carboxyl groups. These acids are corrosive to carbon and low alloy steels,
stainless steels and some nickel alloys at temperature range of 220-420 °C. The presence
of hydrogen sulphide affects the corrosion mechanisms and the following reactions (1-
3) are observed to compete with each other. The last reaction regenerates naphthenic
acid and therefore causes further corrosion. (Ropital, 2009)
Direct acid attack on iron: Fe + 2RCOOH ↔ Fe(RCOO)2 + H2 (1)
Corrosion due to hydrogen sulphide: Fe + H2S ↔ FeS + H2 (2)
Attack of iron naphthenate by H2S: Fe(RCOO)2 + H2S ↔ FeS + 2RCOOH (3)
Petroleum fractions with TAN (Total Acid Number) greater than 1.5 are potentially
corrosive. Besides the TAN, the type of acid and the presence of H2S are important
parameters. For naphthenic acids with high molecular weight, the steric hindrance due
to their long carbon chain would slow the formation of iron naphthenates and hence the
corrosion rate. To prevent the corrosion, steels with increasing chromium and
molybdenum contents offer enhanced corrosion resistance. (Ropital, 2009)
3.7 Calcium
Calcium is usually present in combination with naphthenic acid as soaps (Ali & Abbas,
2006). Offshore crude oil production installations around the world are associated with
the precipitation of calcium naphthenate soaps. The soaps result from the interaction of
polycyclic tetracarboxyl acids with divalent calcium ions (Ca2+) present in produced
waters. These tetraprotic naphthenic acids are known as ARN acids. Due to the four
reactive carboxylate groups within each molecule, the ARN acids are cross-linked by
15
calcium ions to form long polymeric networks. The calcium naphthenate soaps are
insoluble in both crude oil and water. (Juyal et al., 2015)
The precipitation of calcium naphthenate soaps causes significant operational
challenges. Calcium naphthenate scale is one of the most challenging depositions in
crude oil production. The deposits occur as viscous, sticky solids and sludges that harden
upon exposure to air causing physical blockages in production facilities, for example in
oil-water separators, desalters, heat exchangers and pipelines. Moreover, the deposits
need to be physically removed which increases maintenance costs and possibly leads to
replacement of production equipment. (Juyal et al., 2015)
3.8 Silicon
In petroleum products, the presence of silicon is mainly due to the use of antifoaming
agents such as polydimethylsiloxanes (PDMS). The antifoaming agents are used to
enhance the crude oil recovery from the reservoir and to avoid the formation of
emulsions in refinery processes. PDMS degrades around 300 °C generating various
different silicon degradation products, such as siloxanes, silanes and silanols, which can
then react with the hydrocarbon matrix. (Chainet et al., 2011; Chainet et al., 2013)
By adsorbing on catalyst surface, silicon acts as a catalyst poison causing deactivation.
However, silicon species have different effects on the catalytic process depending on the
nature of the catalyst, on the experimental conditions and especially on the chemical
nature of the silicon molecule. Due to the instability of several silicones, the reactivity
and the chemical nature of the silicon compounds are unknown which complicate the
understanding of silicon poisoning. (Chainet et al., 2011; Chainet et al., 2013)
16
4 Impurity removal methods
The aim of this thesis is to study impurity removal methods preceding sulphur removal
unit to decrease the deactivation of desulphurization catalyst. Therefore, the process for
removing sulphur is not discussed. The impurity removal methods presented are divided
into physical and chemical methods.
Metal deposition on hydrotreatment catalysts is irreversible and there is no possibility
of recovering the initial catalytic activity of the catalysts. The pore plugging caused by
metal deposits cannot be removed which sets a limit to the activity recovery. (Toulhoat
& Raybaud, 2013) Thus, finding an effective metal removal method would extend the
catalyst lifetime and enable better activity recovery.
4.1 Physical method
The physical method discussed is an extraction treatment called solvent deasphalting.
An advantage of the solvent deasphalting is that the extraction can be performed under
moderate operating conditions. However, finding a solvent that is sufficiently selective
can be problematic. It is reported that extraction treatment leads to a loss of 5-15 wt%
of the initial amount of saturated hydrocarbons. (Gaile et al., 2001)
4.1.1 Solvent deasphalting
Solvent deasphalting (SDA) is a separation process in which the residue is separated by
molecular weight instead of by boiling point. In the SDA process, heavy feedstock is
separated into deasphalted oil (DAO) with low metal content and a heavy pitch stream
containing most of the contaminants. The feed is mixed with a light paraffinic solvent,
typically C3-C7. Due to their insolubility to these solvents, asphaltenes will precipitate out
of the mixed feedstock. (Fahim et al., 2010) An advantage of SDA extraction is that it can
17
be operated in relatively low temperature and pressure conditions (Lee et al., 2014).
However, the process removes convertible material along with the metal-containing
species (Mandal et al., 2014).
As the metals, sulphur and nitrogen are mainly concentrated in the heavier molecules of
petroleum, their content in DAO is significantly reduced (Parkash, 2003). This can be seen
in Figure 7 where vacuum residue was used as a feed to SDA process.
Figure 7. Percent of demetallization, denitrification and desulphurization as a function of DAO yield (Parkash, 2003).
The SDA process consists of extraction, DAO recovery, asphalt recovery and solvent
recycling system (Parkash, 2003). A process flow chart is presented in Figure 8. Due to
its high viscosity, the feed is mixed with a small amount of solvent prior to the extraction
tower. The feed, with a relatively small amount of solvent, enters the extraction tower
at a point about two-thirds up the column while the solvent enters near the bottom of
the tower. The extraction tower is a multistage contactor equipped with, for example,
18
baffle trays or rotating discs. In the tower, the heavy oil flows downward as the light
solvent goes upward. Asphalt is insoluble in the solvent and therefore, it precipitates and
flows towards the bottom of the tower. (McKetta, 1992; Parkash, 2003)
Figure 8. Process flow chart of a SDA process (McKetta, 1992).
The separation of the asphalt from the oil is controlled by varying the solvent/oil ratio
and by maintaining a temperature gradient across the extraction tower. The highest
temperature is at the top of the tower and the lowest at the solvent inlet. Steam is used
in the heating coils in the top of the tower to regulate the temperature. The steam flow
is adjusted to the feed inlet temperature. (McKetta, 1992)
The dissolved DAO leaves the top of the tower and flows to an evaporator where
vaporized solvent is separated from the DAO. The liquid flows from the bottom of the
19
evaporator to a flash tower where most of the remaining solvent is vaporized. Finally,
the DAO is fed to a steam stripper operating at atmospheric pressure. Superheated
steam enters the lower part of the tower. The remaining solvent is stripped out and flows
overhead with the steam through a condenser into a compressor suction drum where
the water is removed. The DAO product is obtained from the stripper bottom. (McKetta,
1992)
The asphalt and solvent mixture obtained from the bottom of the extraction tower flows
through a heater into a flash drum. In the flash drum, the vaporized solvent is separated
from the asphalt. Asphalt, with a small quantity of solvent, flows from the bottom of the
flash drum to an asphalt steam stripper. The steam and solvent vapors leave from the
top of the stripper and are mixed with the DAO stripper overhead. Solvents recovered
from each of the process steps are recycled into solvent accumulators and can be reused.
(McKetta, 1992) The asphaltene-rich pitch can be utilized to produce heat and hydrogen
by gasification and as a road-packing material (Lee et al., 2014).
The selection of a solvent significantly affects the performance, flexibility and economics
of the process. The solvent must be suitable for the extraction of desired oil fraction and
for the control of the yield and/or the quality of the DAO at temperatures which are
within the practical operating limits. (McKetta, 1992) Selectivity is the ability of the
solvent to separate the paraffinic and possibly resinous oils from the asphalt. The
selectivity can be improved by increasing the solvent/oil ratio at a constant DAO field.
However, there is always an optimum solvent/oil ratio for each operation due to the
energy consumption in the recovering of the solvent. (Parkash, 2003) Increasing the
carbon number of the solvent reduces the quality of the DAO but enhances the yield of
DAO produced because hydrocarbons of higher molecular weight become soluble in the
solvent (Lee et al., 2014).
If the operating temperature approaches the critical temperature of the solvent, the
process becomes unstable. For example, the critical temperature of propane is 97 °C
20
which limits the extraction temperature to about 80 °C. Generally, the greater the
temperature difference between the top and the bottom of the extraction tower, the
better will be the quality of the DAO. This results from the internal reflux that is
generated in the tower. However, too much internal reflux can cause flooding. The
extraction tower pressure must be higher than the vapor pressure of the solvent at the
tower operating temperature. (McKetta, 1992)
4.2 Chemical methods
In this chapter, chemical methods including demetallization processes using acids,
phosphorous compounds and supercritical water are discussed. Also a photochemical
denitrogenation and desulphurization process is presented. Instead of chapter 4.2.1, the
metal removal methods utilizing phosphorous and phosphoric acids are discussed in
chapter 4.2.2 among the other phosphorous compounds.
4.2.1 Demetallization by acids
Generally in a demetallization process with an acid, the metal compounds are converted
into water-soluble constituents or to constituents separable in aqueous phase from the
oil phase (Powell, 1957). The demetallization of oil-soluble metalloporphyrins (MP) by
acids (HX) is a reversible reaction which can be presented by the following equation (Ali
& Abbas, 2006):
MP + HX ↔ PH + MX (4)
Abbas et al. (2010) treated heavy furnace oil with various chemicals to remove nickel and
vanadium. Among the acids used in the experiments, sulphuric acid showed the most
promising results. (Abbas et al., 2010) Moreover, treatment of petroleum fractions with
sulphuric acid to improve the quality has been used commercially for many years. In the
treatment with sulphuric acid, some of the acid is almost always reduced to sulphur
21
dioxide. Consequently, sulphur dioxide may react with unsaturated hydrocarbons
forming additional products such as sulphones, polysulphones and aromatic sulphonic
acids. It is a disadvantage that sulphuric acid reacts with and promotes reactions of
hydrocarbons. (Ali & Abbas, 2006) Another problem in the demetallization process is the
possible formation of a stable emulsion. The more viscous the oil is the more readily it
emulsifies with the acid. (Abbas et al., 2010)
Powell (1957) patented a process for the removal of heavy metals with sulphuric acid.
The process was designed to be prior to a catalytic cracking operation. The sulphuric acid
used in the demetallization had a concentration of 10 % acid. After the oil was mixed
with the dilute sulphuric acid, it was washed with water in order to transfer the acidified
metal components from the oil to the aqueous phase. If no difficult emulsions were
formed, the mixture could be settled to effect the separation of oil and aqueous phases.
In cases where difficult emulsions were produced, the mixture was neutralized with an
alkaline agent to a pH value within 6.5-7.5 before the separation of the aqueous phase
from the oil phase. Another option to resolve the emulsion was to use electrical
precipitation. (Powell, 1957)
Kimberlin Jr. and Judson (1959) patented a technique also utilizing sulphuric acid for the
removal of nickel, vanadium and iron contaminants from heavy gas oil before catalytic
cracking process. Sulphuric acid reacted with the contaminants to form acid sludge which
could be separated from the oil phase by settling/decantation. The optimal processing
temperature was 93 °C at atmospheric pressure with the sulphuric acid concentration of
80 %. Higher temperatures resulted in lower oil yields with high acid consumption due
to the occurrence of oxidation reactions. (Kimberlin Jr. & Judson, 1959)
Various different acids have been studied and patented for the removal of organically-
bound calcium compounds from hydrocarbonaceous feedstocks. Kuehne et al. (2005)
patented a method where the calcium-containing hydrocarbonaceous feedstock was
treated with aqueous extraction solution consisting of acetate ion and alkaline material
22
at the pH range of 3.0-5.0. Acetic acid was a suitable source of acetate ions and, for
example, ammonium hydroxide, sodium hydroxide and potassium hydroxide were
possible alkaline materials. While the alkaline material was added to keep the pH value
within the particular range, the acetate ions of the solution formed water-soluble
complexes with calcium. (Kuehne et al., 2005)
According to the patent, acetate ion at the specific pH facilitated the decomposition of
calcium compounds and provided a mechanism for more easily transporting the calcium
ions from the oil phase to the aqueous phase. Surprisingly high calcium removal in the
pH range between 3.0 and 5.0 was related to the pH at which organically-bound calcium
was most easily dissociated. The calcium removal as a function of the pH value is
presented in Figure 9. The pH range of 3.0-5.0 also appeared to provide a low interfacial
tension between the aqueous and oil phases which facilitated the transport of calcium
into the aqueous phase. (Kuehne et al., 2005)
Figure 9. Calcium removal as a function of the pH value of the extraction solution (Kuehne et al., 2005).
23
After mixing the petroleum feedstock with the extraction solution, the mixture was
separated into aqueous phase and calcium-reduced oil phase. The phases could be
separated by a simple decanting process if no emulsion was formed during the process.
However, an emulsion was often formed and needed to be broken or demulsified before
the separation. The mixing process was generally done at a temperature below the
boiling point of water at the process pressure. Typically, temperatures ranged from 25
to 200 °C and the pressure was greater than the atmospheric pressure. (Kuehne et al.,
2005)
Adams et al. (1965) patented a process for removing metals from high boiling
hydrocarbon fractions with hydrofluoric acid. The hydrofluoric acid reacted with the
metallic contaminants to form finely divided solid particles. The reaction effluent was
treated to remove the acid by, for example, flashing or stripping with steam before the
remaining oil and solid metallic particles were separated. The metals were recovered as
a solid containing 5-20 wt% of metal in a carbonaceous, coke-like material. This metallic
coke-like material was insoluble in both the oil and the aqueous phases while the bulk of
the porphyrin molecule remained in the oil phase. The metallic contaminants could be
removed by filtration. This demetallization process differed from other methods using
acids because the metal contaminants were recovered as solid precipitates rather than
as acid-soluble compounds or heavy sludge. (Adams et al., 1965)
According to the patent of Adams et al., operating the demetallization process
continuously was preferable since no separate layers of acid and hydrocarbons were
allowed to form in a continuous process. A process flow chart containing a distillation
unit of crude oil and the demetallization process is presented in Figure 10. The distillation
unit could be an atmospheric distillation or a combination of atmospheric and vacuum
distillation towers. The hydrofluoric acid could be continuously removed as vapor from
the flash tower and recycled back to the process. (Adams et al., 1965)
24
Figure 10. Distillation unit and demetallization process with hydrofluoric acid (Adams et al., 1965).
The concentration of aqueous hydrofluoric acid used in the process was from 50 % to
nearly 100 %. However, high concentration (70-99 %) was more preferable. The
temperature of the demetallization was preferably between 120 and 180 °C and the
pressure in the mixer needed to be sufficient to maintain the reactants in liquid phase at
the temperature used. The oil and acid could be intensively agitated to obtain a good
contact since the formation of emulsions between the oil and the acid was not a problem
as the separation was done by flashing. After mixing, the effluent mixture passed through
a pressure relief valve to the flash tower where the acid vapors were flashed overhead
25
and condensed before recycling. The process could recover a hydrocarbon yield of over
99 %. (Adams et al., 1965)
4.2.2 Demetallization by phosphorous compounds
Kukes and Battiste (1985) patented a demetallization process which utilized
phosphorous acid. Phosphorous acid reacted with the metals to form oil-insoluble
compounds that could be removed from the oil stream by any conventional method such
as filtration, centrifugation or settling/decantation. Phosphorous acid (H3PO3) was a
more effective demetallizing agent than phosphoric acid (H3PO4) and was found to be
particularly effective in vanadium removal. (Kukes & Battiste, 1985)
The phosphorous acid used in the process was an aqueous solution containing about 40-
80 wt% of phosphorous acid. A gas could be present during the mixing of the
hydrocarbon stream and the phosphorous acid as the gas enabled high-pressure
operation to be achieved. Gases such as hydrogen, nitrogen, air, methane and carbon
dioxide were suitable for the process. When hydrogen was utilized, the pressure of the
demetallization process was preferably 800-7000 kPa. The most preferable temperature
range for the process was between 350-450 °C. According to the patent, the
demetallization process with phosphorous acid was a suitable pretreatment step before
hydrodesulphurization. (Kukes & Battiste, 1985)
A demetallization process in which the hydrocarbon feedstock was contacted with a
phosphorous compound to form oil-insoluble compounds was also patented by Kukes
(1985). Phosphine, hydrocarbylphosphines, hydrocarbylphosphites,
hydrocarbylphosphates, hydrocarbylphosphonates, hydrocarbylphosphine oxides,
hydrocarbylthiophosphites and hydrocarbylphosphine sulfides were suitable
phosphorous compounds for the process. These compounds reacted with the metals
present in the hydrocarbon feed to form oil-insoluble compounds that could be removed
from the hydrocarbon feed by conventional method such as filtration, centrifugation or
26
decantation. This demetallization process was particularly applicable for the removal of
vanadium and nickel. (Kukes, 1985)
In the demetallization process of Kukes, mixing hydrogen with the mixture of
hydrocarbon feed stream and the phosphorous compound was found to enhance the
metal removal. Adding the gas allowed operating the process in high pressure. Compared
to gases such as nitrogen, methane and carbon dioxide, the utilization of hydrogen had
desirable effects on the process such as reduced coking. (Kukes, 1985)
According to the patent, temperature was preferably between 300-450 °C and the
pressure 800-17300 kPa. If the phosphorous compounds were gaseous or liquid, they
could be pumped in that form into the hydrocarbon feed. In the case the phosphorous
compound was solid, it could be dissolved in the hydrocarbon feed stream or in any
suitable solvent. Preferably, the concentration of the phosphorous compound was in the
range of about 0.05 to about 5 wt% based on the weight of the hydrocarbon feed stream.
If an excess of the phosphorous compound was used, the excess phosphorous compound
was removed from the hydrocarbon stream by distillation if the volatility of the
phosphorous compound was suitable. If distillation could not be utilized, the excess
phosphorous compound could be thermally decomposed which generally converted it
to oil-insoluble form which was then removed simultaneously with the metal
contaminants. (Kukes, 1985)
Krambeck et al. (1987) found that water-soluble phosphorous compounds were capable
of forming a compound or a complex with vanadium and nickel present in petroleum. As
a result, a significant quantity of the metal could be removed from the oil phase to the
aqueous phase. They patented a demetallization process in which the
hydrocarbonaceous feedstock was extracted with an aqueous solution of phosphorous
compound. The hydrocarbonaceous feedstock was contacted with a relatively
substantial quantity of water in which one or more phosphorous compounds were
dissolved. The phosphorous compounds formed compounds or complexes with the
27
vanadium and nickel components present in the oil feed and extracted the metals into
the aqueous phase. Examples of suitable and effective phosphorous compounds were
P2O5, H3PO4, (NH4)3PO4, (NH4)2HPO4, (NH4)H2PO4, (NH4)H2P2O7, H4P2O7, H3PO2 and H3PO3.
(Krambeck et al., 1987)
The hydrocarbon stream could be contacted with the aqueous phosphorous solution by
continuous countercurrent extraction. The hydrocarbon stream was fed into the bottom
of the extraction tower while the aqueous phosphorous stream was introduced to the
top of the tower as shown in Figure 11. In the extraction tower, the phosphorous
compounds reacted or formed complexes with the vanadium and nickel present in the
oil. As a result, these metals were extracted from the oil phase into the aqueous phase
which was withdrawn from the bottom of the extraction tower. The phosphorous-metal
compounds/complexes could be separated from the water and therefore, the water
could be recycled back to the process. (Krambeck et al., 1987)
Figure 11. Simplified process flow diagram of the demetallization process by countercurrent extraction (Krambeck et al., 1987).
28
Temperature in the extraction was preferably from about 80 °C to about 200 °C. If the
temperature was above 100 °C, pressurized vessels were required to maintain a liquid
system. However, if the densities of the aqueous phosphorous solution and the oil were
very close and the interfacial tension low, continuous countercurrent extraction was not
applicable and some other contacting procedure needed to be utilized. (Krambeck et al.,
1987)
4.2.3 Demetallization by supercritical water
Utilizing supercritical water (SCW) provides a new approach for the removal of metals
from heavy oils. Metal removal is done above the critical point of water in the absence
of catalyst and without addition of hydrogen. SCW functions as a solvent and a hydrogen
donor. (Mandal et al., 2012)
The properties of SCW differ significantly from the properties of water at ambient
temperature. The dielectric constant of water is 78.5 at 25 °C. The value drops with
increasing temperature and at the critical point of water the dielectric constant is around
6. The relatively low value results from the reduced number of hydrogen bonds. Due to
the low dielectric constant, nonpolar organic substances and gases can be dissolved in
SCW. (Bröll et al., 1999)
The purpose of using SCW is to break the nitrogen-metal bonds of the metal porphyrins.
The demetallization occurs at high temperatures and pressures as the critical point of
water is at 374 °C and 22.1 MPa. The effect of temperature on reaction of metal
porphyrins in SCW is remarkably high while the water partial pressure has a moderate
effect. The metal porphyrin conversion increases with reaction time and higher
temperature as can be seen in Figure 12. However, the reaction did not complete at the
high temperature of 490 °C and the conversion did not change significantly after 90
minutes of reaction time. This indicates the occurrence of a reversible reaction and
establishment of the equilibrium after 90 min at 490 °C. In the experiments, the water
29
partial pressure was 25 MPa and toluene was added as a co-solvent with SCW. Solid
metal porphyrins are stable below 400 °C with water partial pressure of 25 MPa in the
absence of toluene. (Mandal et al., 2014)
Figure 12. Conversion of vanadyl etioporphyrin (VO-EP) and nickel etioporphyrin (Ni-EP) at different temperatures as a function of reaction time (Mandal et al., 2014).
Mandal et al. studied the effect of water partial pressure between 25-45 MPa at 450 °C
in a toluene environment. The conversion of metal porphyrins was found to increase
with higher water partial pressure. The conversion of vanadyl etioporphyrin and nickel
etioporphyrin as a function of water partial pressure is shown in Figure 13. (Mandal et
al., 2014)
30
Figure 13. Conversion of vanadyl etioporphyrin and nickel etioporphyrin as a function of water partial pressure (Mandal et al., 2014).
Mandal et al. used an 8.8 mL batch reactor in their laboratory experiments. The reactor
was loaded to an electric furnace which was heated to the planned temperature. The
reactor was shaken in a front-back motion and after a specific reaction time it was
removed from the furnace and quenched in an ice bath. The water was separated using
a separating funnel and the reaction products were collected by washing with xylene.
(Mandal et al., 2012; Mandal et al., 2014)
The central metal group was found to stabilize the porphyrin molecule with respect to
hydrogenation and ring fragmentation (Bonné et al., 2001). Metal porphyrins are
hydrogenated several times successively until the porphyrin macromolecule is disrupted
and it loses the porphyrinic character. SCW acts as a hydrogen donor in the
hydrogenation reactions. When the covalent metal-nitrogen bonds are weakened, the
porphyrin macromolecule may become more vulnerable to hydrogenation reactions and
ring fragmentation caused by, for example, thermal cleavages. Metal is removed from
the porphyrin structure by reacting with the OH of SCW and by ring fragmentation due
31
to the instability of metal-free porphyrins under the reaction conditions. (Mandal et al.,
2014)
The reaction mechanism could be explained using free radical mechanism. The reaction
was initiated by producing H, OH and metal porphyrin radicals and propagated by
forming radicals of hydrogenated compounds and also metal hydroxide and porphyrin
radicals. Termination occurred when free radicals combined to form stable products. The
mechanism is shown in the following equations. (Mandal et al., 2014)
Initiation:
H2O → H ∙ + ∙ OH (5)
MP ↔ MP ∙ (6)
Propagation:
MP ∙ + 2H ∙ ↔ IHC ∙ (7)
MP ∙ + ∙ OH → M(OH) ∙ + P ∙ (8)
Termination:
P ∙ → Stable ring fragmented product (9)
M(OH) ∙ + ∙ OH → Inorganic metal (10)
IHC ∙ + H ∙ + ∙ OH → Inorganic metal + ring fragmented product + H ∙ (11)
H ∙ + ∙ OH → H2O (12)
Symbols: MP, metal porphyrin; IHC, intermediate hydrogenated compound; M, metal.
However, the fate of the central metal group in the reaction of metal porphyrins under
SCW conditions was obscure. The formation of metal hydroxides, metal oxides and metal
ion was probable. After removing the metal, the porphyrin structure became very
unstable under the experimental conditions and decomposed into light hydrocarbons.
(Mandal et al., 2014)
32
Supercritical water can remove metals from metal porphyrins without the addition of a
catalyst when the temperature is above 400 °C and the water partial pressure is 25 MPa.
The metal removal increased nearly exponentially with the conversion at all
temperatures. At the highest conversion level, approximately 80.3 % of vanadium and
76.4 % of nickel were removed. The SCW based demetallization process has the potential
to remove metals, such as nickel and vanadium, from the metal porphyrins. However,
additional study is required to understand the fate of the metals after the reactions
before the process can be made industrially viable. (Mandal et al., 2014)
4.2.4 Photochemical denitrogenation and desulphurization
Nowadays, the removal of nitrogen from VGO is carried out via a catalytic
hydrodenitrogenation and simultaneous hydrodesulphurization process. However, this
process requires rather severe conditions, utilization of hydrogen and active catalysts.
An alternative process for removing the nitrogen and sulphur compounds
simultaneously under moderate conditions and without the need of hydrogen and a
catalyst has attracted interest. (Shiraishi et al., 2001)
A photochemical denitrogenation and desulphurization process is based on a
combination of UV irradiation and liquid-liquid extraction. Two different solvents have
been applied for the process: water and acetonitrile. Using oil/water system, the
photodecomposition of the sulphur compounds is achieved in the oil phase. This is
followed by the transfer of the resulting decomposition compounds into the water
phase. In the oil/acetonitrile system, the sulphur compounds are extracted into the
acetonitrile phase and there photodecomposed. It is reported that the
photodecomposition of nitrogen compounds proceeds faster than that of the sulphur
compounds by sunlight irradiation in water. Therefore, the denitrogenation may be
performed simultaneously with the desulphurization. (Shiraishi et al., 2000)
33
Shiraishi et al. studied both oil/water and oil/acetonitrile systems. A high-pressure
mercury lamp was used for producing photoirradiation. In their experiments, Shirashi et
al. used model oil containing indole, aniline and carbazole and compared the results with
results obtained from three actual light oils. (Shiraishi et al., 2000)
In oil/water system, indole and aniline photodecomposed faster than carbazole which
decomposition was hindered by the presence of double-ring aromatic hydrocarbons,
naphthalenes. The denitrogenation yield decreased linearly with increasing naphthalene
concentration. However, the denitrogenation rate was found to accelerate by adding
hydrogen peroxide. The rate of all model nitrogen compounds increased with increasing
H2O2 concentration. Hydrogen peroxide affects in two different ways: producing
hydroxyl radicals and acting as a weak oxidizing agent. Direct photodecomposition of
H2O2 produces hydroxyl radicals which react with the nitrogen compounds. As a weak
oxidizing agent, H2O2 oxidizes the photoexcited nitrogen compounds, which also
accelerates the rate of photodecomposition. The photodecomposed nitrogen
compounds were quickly removed from the oil phase into the water where they were
found to appear as NO3- ions. (Shiraishi et al., 2000)
In oil/acetonitrile system, the extraction yield of the nitrogen-containing compounds was
greater than that of the sulphur-containing compounds due to the higher polarity of
nitrogen. After dissolving in acetonitrile, they were photoirradiated. Similarly as in the
oil/water system, the photodecomposed nitrogen compounds were as NO3- ions. In
oil/acetonitrile system, the presence of naphthalene did not decrease the
photoreactivity of nitrogen compounds. (Shiraishi et al., 2000)
Shiraishi et al. were able to reduce the nitrogen content to less than 20 % of the feed
concentration using the oil/water system. However using acetonitrile, the nitrogen
content was successfully decreased to less than 3 % of the feed concentration. Both
systems enabled simultaneous removal of nitrogen and sulphur. (Shiraishi et al., 2000)
34
After experimenting with light oils, Shiraishi et al. studied denitrogenation and
desulphurization of VGO using oil/acetonitrile system. Due to its high viscosity at room
temperature, VGO was diluted with an n-decane solvent at various volume ratios. The
VGO/decane solutions were mixed with acetonitrile and then photoirradiated. The
photoirradiation was done under atmospheric pressure and the temperature of the
mixture was about 50 °C. The nitrogen concentration in the VGO decreased already when
contacted with acetonitrile but it decreased dramatically during photoirradiation.
Therefore, the nitrogen compounds efficiently photodecomposed in acetonitrile into
highly polarized compounds such as NO3- ion. As can be seen in Figure 14, the
denitrogenation rate was found to increase when VGO/decane and VGO
solution/acetonitrile volume ratios decreased. The data points for zero irradiation time
are those obtained by simply mixing the two phases. (Shiraishi et al., 2001)
35
Figure 14. The decrease in the total nitrogen content in VGO during photoirradiation of an oil/acetonitrile system using different VGO/decane and oil/acetonitrile volume ratios (Shiraishi et al., 2001).
In the experiments, the photoirradiation time was 10 hours. The nitrogen content of VGO
was decreased to less than 1 % of the feed concentration when the oil/acetonitrile and
VGO/decane volume ratios were 1/5 and 1/9, respectively. Similarly, the
36
desulphurization rate was found to increase when VGO/decane and oil/acetonitrile
volume ratios decreased. As a result, also the sulphur content was decreased to less than
1 % of the feed concentration. (Shiraishi et al., 2001)
5 Advantages and disadvantages of the methods
In this chapter, the advantages and disadvantages of the impurity removal methods
presented in chapter 4 are discussed and the methods are compared to each other. One
process is chosen for prestudy which is carried out in the applied part of this thesis. The
reasoning behind the selection is presented.
5.1 Comparison of the methods
The temperatures, pressures, solvents and reagents used in the SDA, demetallization
processes by acids, phosphorous compounds and SCW as well as in photochemical
denitrogenation are summarized in Table 2. The photochemical denitrogenation has the
mildest operating conditions while the process utilizing SCW requires the most extreme
conditions. Due to the high temperature and pressure in the SCW process, the
hydrocarbons may start to decompose during the process.
37
Table 2. Temperatures, pressures, solvents and reagents used in each impurity removal process.
Impurity removal process
Temperature (°C)
Pressure (kPa)
Solvent/reagent
SDA 80–250 2000–4000 light paraffinic solvent (C3-C7)
Demetallization by acids
25–200 101–2000 H2SO4, CH3COOH + alkaline material, HF
Demetallization by phosphorous compounds
100–450 101–17500 various suitable phosphorous compounds including phosphorous acid, in certain processes also hydrogen or water
Demetallization by SCW
> 400 > 25000 water + toluene
Photochemical denitrogenation
50 101 n-decane + water or acetonitrile
The National Fire Protection Association (NFPA) has developed a system for identifying
hazards associated with materials. The system provides basic information for emergency
personnel responding to a fire or spill and for those planning the emergency response. It
indicates the health hazard, flammability, reactivity and special hazards for many
common chemicals. The hazards are rated from 0 to 4 where 0 indicates the least
hazardous and 4 the most hazardous. (NFPA, 2012) The reactivity of a chemical describes
the reactivity with water and the capability of detonation or of explosive reactions. The
special hazards indicate if a chemical functions as an oxidizing agent (OX) or if it shows
unusual reactivity with water (W). (Northeastern University, 2015) The hazard ratings
available for the chemicals used in the impurity removal processes are presented in Table
3.
38
Table 3. NFPA ratings available to chemicals used in the impurity removal processes (Northeastern University, 2015).
NFPA hazard rating system
Impurity removal process Chemical Health Flammability Reactivity
Special information
SDA Propane 1 4 0
Butane 1 4 0
Pentane 1 4 0
Hexane 1 3 0
Heptane 1 3 0
Demetallization by acids
H2SO4 3 0 2 W
CH3COOH (glacial) 2 2 2
NaOH 3 0 1
KOH 3 0 1
HF 4 0 0
Demetallization by phosphorous compounds
Phosphine 3 4 1
H3PO4 2 0 0
Hydrogen 0 4 0
Demetallization by SCW
Toluene 2 3 0
Photochemical denitrogenation
Decane 0 2 0
Acetonitrile 2 3 0
W: reacts strongly with water
The NFPA hazard rating system gives valuable information about the chemicals and their
properties. By utilizing chemicals with low ratings, the hazards of a process can be
reduced. Unless a certain chemical is critical for the process, replacing it with a less
hazardous alternative is preferable.
Hydrofluoric acid has the most hazardous rating (4) for the health. It is caustic to
respiratory tract and toxic if inhaled. It causes severe skin burns and eye damage and
may cause frostbite. (Airgas, 2015) Moreover as presented in chapter 4.2.1, the
hydrofluoric acid preferred in the demetallization process has a really high concentration
(70-99 %). Therefore, preferring the use of other chemicals could reduce the health risks.
39
As can be seen in Table 3, light paraffinic hydrocarbons, hydrogen and phosphine are
extremely flammable. They pose a threat of causing an explosion. Therefore, their
utilization and handling requires special caution in order to safely operate the process. If
the risk of explosion cannot be minimized to an acceptable level and the chemicals
handled safely, the process is not implementable.
The properties of chemicals affect the material choices as the chemicals can cause
corrosion. The main parameters affecting the acid corrosion are temperature, acid
concentration, the hydrodynamic flow rate and the presence of oxidizing agents.
Formation of, for example, metal sulfate or fluoride layer protects the alloys but if the
layer is torn, serious corrosion will occur. The presence of oxidizing agents increases the
corrosion rate. (Ropital, 2009) For example for the handling of sulphuric acid, austenitic
stainless steels have a good resistance only up to 65 °C over the entire concentration
range. Some nickel-molybdenum alloys can be used even in temperatures above 120 °C.
However, these alloys suffer serious corrosion in the presence of oxidizing agents and in
aqueous acid systems. The corrosion rates of nickel-base alloys generally increase with
acid concentration up to 90 % but higher concentrations are usually less corrosive.
(Davis, 2000) Compromises between the operating conditions and acid concentration
might need to be done in order to find suitable materials.
The materials used in SCW applications must simultaneously be resistant to high
temperatures and pressures and also to corrosion by substances employed (Bröll et al.,
1999). The most commonly used reactor materials are stainless steels and nickel-base
alloys. The alloying elements have to form an oxide layer which completely covers the
alloy. Chromium as an alloying element improves the resistance against acidic and
oxidizing media while nickel improves the corrosion resistance in alkaline environments.
However, chromium loses its protective effect with increasing temperature due to
chromate formation. Chromate formation typically occurs when temperature is above
200-250 °C resulting in high corrosion rate. Anions can also destroy the oxide film or act
40
as oxidizing agents corroding the metals. More basic physical and chemical data is
required and extensive long-time studies need to be performed to gain more information
about the corrosion resistance of materials in SCW. (Kritzer, 2004)
The demetallization process by SCW and the photochemical denitrogenation are novel
processes. Therefore, they are not as technically mature as SDA and demetallization by
acids and phosphorous compounds. The lack of technical maturity is a disadvantage of
these processes as they require more research and development before they can be
operated in an industrial scale. Moreover, as finding suitable materials for SCW is
problematic, the demetallization process utilizing SCW is currently not feasible.
However, utilizing water as the reagent is an interesting option and subject of research
as it could eliminate the need of using toxic and flammable chemicals.
Equipment required for the impurity removal in the more technically mature processes
is presented in Table 4. Depending on formation of an emulsion, the demetallization
processes by acids may require a unit for breaking or demulsifying the emulsion.
Generally, the more equipment is required in a process, the more expensive the
investment is. Moreover, the amount of process equipment affects the maintenance
costs and the required land area. As the new impurity removal process would be a part
of an existing refinery production line, the area available for construction is limited. If
the area close to the existing production line is too small, the new process could be
constructed slightly further increasing the pipeline and pumping costs.
41
Table 4. Equipment required in different impurity removal processes.
Impurity removal process Equipment for removing impurities
SDA Extraction tower, 2 flash towers, 2 strippers
Demetallization by acids Mixer, decanter
Mixer, flash tower or stripper, filter
Demetallization by phosphorous compounds
Mixer, decanter or filter, possibly a distillation tower for the removal of excess phosphorous compounds
Extraction tower, water separator
An advantage of SDA is the possibility to recycle the solvent. Because SDA is a physical
method, the impurity removal is not based on chemical reactions and the separation of
solvent is straightforward using flash towers and strippers. In the experiments of
photochemical denitrogenation and desulphurization, the removal of nitrogen and
sulphur was found to increase when the amount of solvent increased. High impurity
removal rate was achieved using significant quantities of solvent compared to the
amount of oil treated. Consequently, recycling the substantial amount of solvent could
be challenging and cost-inefficient. If the solvent cannot be reused, the process would
require an excessive fresh solvent stream which would not be feasible.
Furthermore, the photochemical denitrogenation and desulphurization process is
designed for the removal of nitrogen and sulphur impurities while the other methods
presented in chapter 4 are for the demetallization of VGO. SDA removes simultaneously
the asphaltenes and metals. As the objective of the photochemical denitrogenation and
desulphurization differs from that of the other methods, the photochemical process
cannot be directly compared to the demetallization processes. Depending on the
impurity levels of VGO, denitrogenation would be more interesting if the metal
concentration in the feedstock was low but the nitrogen and sulphur content high.
42
5.2 Selection of one process for prestudy
The most interesting methods for the impurity removal of VGO are SDA and
demetallization using phosphorous compounds. The possibilities of SDA have already
been noticed in Neste and the process has been studied for years. Currently, a SDA
process unit is under construction in Porvoo refinery. However, the SDA will be used for
heavier feedstock than VGO to reduce the asphaltene content in the feed. The process
can reduce the asphaltene and metal concentrations but not completely remove those
impurities. VGO may be too light to achieve a significant and cost-effective reduction in
the impurity levels.
The patents made of the demetallization processes with phosphorous compounds were
published in the 1980s and thus, they have already expired. In addition, no licenses for
the demetallization process were found. Therefore, the process could be freely designed.
Compared to SDA, the applicability of this demetallization method has not yet been
studied and choosing this for the prestudy would provide new information.
A valuable outcome of this thesis is providing new information about the impurity
removal systems. Therefore, the demetallization process by phosphorous compounds
was chosen for prestudy. The demetallization process will be applied as a pretreatment
process before hydrodesulphurization of VGO in order to reduce the deactivation of
hydrotreatment catalyst. The prestudy is carried out in the applied part of the thesis.
43
APPLIED PART
6 Designing of an impurity removal unit
The amount of VGO fed to the HDS process is 300 tons/h and it contains 10 ppm of metal
impurities. VGO with 10 ppm of metal impurities is applied in this work as an example of
a low-quality VGO. The aim is to reduce the amount of metal impurities by applying a
pretreatment process which utilizes phosphoric acid. The demetallization process
extends the HDS catalyst lifetime and therefore, the catalyst can be changed less
frequently.
The phosphorous compound used in the demetallization process must be safe to operate
and available in bulk quantities. Therefore, phosphoric acid was chosen as the
demetallizing agent over phosphorous acid. Phosphorous acid is an unstable substance
which converts to phosphoric acid and phosphine when heated. The decomposition of
phosphorous acid is presented in Equation 13. (Pauling, 1970) The formed phosphine
(PH3) is extremely flammable. Its autoignition temperature is 38-100 °C depending on
the concentration and diluent. (Pohanish, 2012) Due to the formation of phosphine, the
utilization of phosphoric acid in the demetallization process is preferred over
phosphorous acid.
4H3PO3 → 3H3PO4 + PH3 (13)
The amount of experimental data about the metal impurity removal capacity of H3PO4 is
limited. The data found is presented in Table 5. As can be seen, the impurities studied
are often limited to only nickel and vanadium. The metal removal capacities of Eidem
(1988) also include the removal of iron and therefore, in this applied part of the thesis
the metal impurities, the concentrations of which are reduced by H3PO4, are nickel,
44
vanadium and iron. The concentrations of the other metal impurities present in VGO are
not supposed to be affected by H3PO4 as no data could be found about their removal
rate. To obtain more accurate data for the demetallization of VGO, laboratory
experiments should be performed. The amount of H3PO4, operating temperature and
the initial concentration of impurities affect the removal capacity of H3PO4 and
therefore, laboratory experiments would provide valuable information for the process
design.
Table 5. Experimental data of the metal removal capacity of H3PO4.
Removal of impurities
(weight-%)
Acid T (°C) Weight-% of acid
in the mixture V Ni Fe Source
H3PO4, 17% 95 29 35 37 NA (Krambeck et al., 1987)
H3PO4, concentrated
250 5 18 10 NA (Abbas et al., 2010)
H3PO4, 85.1% 260 1 23.4 32.1 21.7 (Eidem, 1988)
H3PO4, 85% 400 1.2 60.0 4.1 NA (Kukes & Battiste, 1985)
H3PO4, 85% 400 3.8 41.5 23.5 NA (Kukes & Battiste, 1985)
NA (not available)
The concentrations of metal impurities in the VGO feed are listed in Table 6. The
concentrations are determined based on average concentrations of actual VGO feeds
used in the refinery. However, the values are roughly rounded and therefore, do not
directly represent the concentration of any actual VGO feed. The metal removal
capacities of Eidem (1988) are used in the calculations as the experiments include the
removal of iron and the removal percentages are not the highest or the lowest compared
to the other publications. Therefore, the calculations would not give either too optimistic
45
or too poor results but an average removal of the metal impurities. Based on the data,
the removal percentages of vanadium, nickel and iron used in the calculations are 23 w-
%, 32 w-% and 22 w-%, respectively. To determine the removal percentages more
accurately for this demetallization process, laboratory experiments are necessary. The
mass flow rates of the metal impurities in the feed and in the demetallized VGO are also
presented in Table 6.
Table 6. Concentrations and mass flow rates of metal impurities in the VGO feed and mass flow rates after the demetallization process.
Impurity Concentration in VGO
feed (ppm) Mass flow rate in the
feed (kg/h) Mass flow rate in VGO after
demetallization (kg/h)
Ni 3 0.90 0.61
V 3 0.90 0.69
Na 2 0.60 0.60
Fe 1 0.30 0.23
Ca 0.9 0.27 0.27
As 0.1 0.03 0.03
Total 10 3.00 2.44
The applied part of the thesis is divided into case 1 and case 2. In case 1 operating the
HDS process without an impurity removal unit is studied and the HDS catalyst lifetime is
calculated. In case 2 the demetallization process is integrated into the HDS process. The
demetallization process is designed, the equipment sizing is performed and the costs of
implementing this new impurity removal treatment are estimated. Finally, the
profitability of the demetallization process is assessed.
46
7 Case 1. HDS process without a new impurity removal system
In this chapter, the HDS process without the demetallization pretreatment is discussed.
The HDS catalyst lifetime is calculated in order to determine how fast the VGO feed with
10 ppm of metal impurities deactivate the catalyst.
7.1 Process description
HDS is the primary sulphur removal technology used in refineries. It is a catalytic process
in which the sulphur content in fuels is reduced by reaction with hydrogen at 330-370
°C. (Shekhawat et al., 2011) As a result of the reaction, H2S is formed. Molybdenum is
often used to formulate HDS catalysts but tungsten is also effective. These metals are
used in combination with nickel or cobalt which acts as a promoter. (Schobert, 2013)
Hydrogen sulphide produced during the HDS must be treated and therefore, the H2S/H2
stream is fed to an amine scrubber to absorb the H2S. After the amine scrubbing, H2S is
desorbed and sent to a sulphur recovery process. (Shekhawat et al., 2011) However, in
this thesis the focus is in the beginning of the HDS process and in the reactors. Therefore,
the processing of reactor effluent is excluded from this work.
Process flow chart of the examined part of HDS process is presented in Figure 15. The
feed is pumped from a storage tank (FA-101) through two heat exchangers (EA-101 and
EA-102). In the heat exchangers, the effluent of the last reactor (DC-103) is utilized to
heat the feed stream. To raise the temperature even higher, the feed flows to a furnace
(BA-101). Then, the hydrodesulphurization occurs in three consecutive reactors (DC-
101…103). Hydrogen is added to the feed before the furnace and in the first reactor.
47
Figure 15. Process flow chart of the examined part of HDS process.
7.2 Catalyst lifetime
The HDS catalyst is considered to be deactivated when it has 30 % left of the initial
activity. Guard bed is utilized to adsorb the metal impurities and to limit the deactivation
of HDS catalyst. Based on the experience in Neste, the guard bed is estimated to adsorb
25 w% of the metal impurities in the feed until its adsorption capacity is full. The total
adsorption capacity is around 10 % of the mass of guard bed. As the mass of guard bed
catalyst used in the process is around 12 tons, the adsorption capacity of the guard bed
48
is 1.2 tons of metal impurities. The amount of different metal impurities adsorbed by the
guard bed is presented in Table 7. The process is started in week 1 and it can be seen
that the adsorption capacity is full during week 10. As a result, the guard bed cannot
adsorb more and would let all the metal impurities go through to the HDS catalyst after
week 10.
Table 7. Accumulation of metal impurities in guard bed.
Accumulation of impurities in the guard bed (kg)
Week
Feed stream
(t/h)
Feed stream (t/wk) As Ca Fe Na V Ni Total
1 300 50400 1.3 11.3 12.6 25.2 37.8 37.8 126
2 300 50400 2.5 22.7 25.2 50.4 75.6 75.6 252
3 300 50400 3.8 34.0 37.8 75.6 113.4 113.4 378
4 300 50400 5.0 45.4 50.4 100.8 151.2 151.2 504
5 300 50400 6.3 56.7 63.0 126.0 189.0 189.0 630
6 300 50400 7.6 68.0 75.6 151.2 226.8 226.8 756
7 300 50400 8.8 79.4 88.2 176.4 264.6 264.6 882
8 300 50400 10.1 90.7 100.8 201.6 302.4 302.4 1008
9 300 50400 11.3 102.1 113.4 226.8 340.2 340.2 1134
10 300 50400 12.6 113.4 126.0 252.0 378.0 378.0 1260
The deactivation of the HDS catalyst is calculated using research data of Neste. Based on
the data, the activation of the HDS catalyst will go below 30 % of the initial activity during
week 9. Therefore, the HDS catalyst lifetime is 8 weeks when VGO feed with 10 ppm of
metal impurities is used. The HDS catalyst is deactivated before the adsorption capacity
of the guard bed is full. Due to the short catalyst lifetime, the process is shut down after
every 8 operating weeks resulting in lost profit. Both the HDS and the guard bed catalyst
are changed during the shutdowns. As the shutdowns are required often, the process is
under normal operating conditions only for short periods of time. Therefore, the process
risks are higher as the most of process deviations occur when the process is being shut
down or started after maintenance.
49
By reducing the metal impurity level before the HDS, the catalyst lifetime could be
extended and the shutdowns would occur less frequently. The costs of changing the HDS
catalyst and the guard bed during one shutdown are equal in cases 1 and 2. The costs of
changing the catalysts are discussed in chapter 8.3.2.
8 Case 2. Impurity removal integrated in the HDS process
In this chapter, the demetallization process by phosphoric acid is designed and the sizing
of required equipment is performed. The profitability of the process is evaluated and in
the end of this chapter, the focus is on safety issues.
8.1 Process design
The demetallization with phosphorous compounds can be realized using two different
approaches. The first process option is countercurrent extraction which was described
in the patent of Krambeck et al. (1987). The other option is to use a mixer followed by a
decanter or filter (Kukes, 1985).
The extraction requires significant amount of water. According to Krambeck et al. (1987),
the weight ratio of water to oil was most preferably 0.2-1 meaning 60-300 t/h of water
in the demetallization process. Therefore, the capacity of handling, cleaning and
recycling the water should be enormous. Due to the large amount of water required in
the extraction, using a mixer was preferred as the process option.
8.1.1 Process flow chart
The mixing is done in a static mixer, a tubular device which consists of stationary mixing
elements. The mixing elements split the fluid flow into several partial flows which are
then recombined to form a mixture. Both the degree of homogeneity and the pressure
50
drop increase with the number of mixing elements. Because of no moving parts, they
require little or no maintenance. Another advantages are short residence times and that
there are no power requirements other than pumping the fluid. (Coker, 2001) After the
mixer, the oil and aqueous phases are separated. For the separation, both a decanter
and an electrostatic coalescer are studied but only the decanter is designed.
The demetallization process was chosen to be located between the two heat exchangers
(EA-101 and EA-102) of HDS process. Process flow chart of the HDS with the integrated
demetallization is presented in Figure 16. VGO feed flows through the heat exchanger
EA-101 before it enters the static mixer (GD-101) where phosphoric acid is added to the
VGO stream. The operating temperature is around 225 °C and the pressure 60 bar in the
mixer. Then, the mixture flows to either a decanter or an electrostatic coalescer (HB-101)
where the oil phase is removed from the aqueous phase and fed to the furnace (BA-101)
through the heat exchanger EA-102. Finally, the demetallized VGO enters the HDS
reactors (DC-101…103).
51
Figure 16. Process flow chart of demetallization process integrated in HDS process.
Even though the demetallization process reduces the amount of metal impurities
entering the reactors, the same guard bed will be used in this process as in case 1. The
guard bed further extends the HDS catalyst lifetime.
8.1.2 Mass flow rate of phosphoric acid
The amount of H3PO4 required in the mixer is calculated based on reactions of the metal
impurities with phosphate ions. In petroleum nickel occurs in the +2 valence state while
most vanadium atoms have a valence of +4, almost exclusively as vanadium ions VO2+
(Mandal et al., 2014). Similarly to nickel and vanadium, iron is present in the structures
of porphyrin compounds (Leprince, 2001). The reactions with phosphate ions are
presented in Equations 14-16.
52
3Ni2+ + 2PO43− ↔ (Ni)3(PO4)2 (14)
3VO2+ + 2PO43− ↔ (VO)3(PO4)2 (15)
3Fe2+ + 2PO43− ↔ (Fe)3(PO4)2 (16)
Based on the equations, the required molar amount of phosphoric acid is
n(H3PO4) =2
3n(Ni) +
2
3n(V) +
2
3n(Fe) (17)
The molar flow rates of nickel, vanadium and iron in the VGO feed are presented in Table
8. Utilizing large excess of H3PO4 would require more separation capacity for the
decanter as the unreacted H3PO4 should be removed. However, a small excess of
phosphoric compound is necessary in order to remove all the possible metal impurities
in a reasonable time. Therefore, the preliminarily estimated amount of H3PO4 used in the
process is twice the stoichiometric molar amount.
Table 8. Mass flow, molar mass and molar flow of nickel, vanadium and iron in the VGO feed.
Impurity Concentration (ppm) m (kg/h) M (g/mol)* n (mol/h)
Ni 3 0.9 58.693 15.33
V 3 0.9 50.942 17.67
Fe 1 0.3 55.845 5.37
*Source: (Yaws, 2011)
Based on Equation 17, the stoichiometric molar flow of H3PO4 is 25.6 mol/h. Therefore,
the molar amount applied in the process is 51.2 mol/h. Based on the molar masses
published by Yaws (2011), the molar mass of H3PO4 is 97.996 g/mol and therefore, the
mass flow of 100 % H3PO4 is 5.0 kg/h. However, H3PO4 is corrosive and the use of 100 %
53
H3PO4 is not required in the demetallization process. Therefore, an aqueous solution
containing 75 w% H3PO4 is chosen to be utilized in the process. The mass flow of 75 w%
H3PO4 is 6.7 kg/h which is rounded to 7 kg/h. The mass flow of H3PO4 is rounded to 7
kg/h as the mass flow of VGO is also expressed with one significant number within the
accuracy of the thesis.
8.1.3 Physical properties of VGO and phosphoric acid
Physical properties of VGO used in the calculations are listed in Table 9. The density and
viscosity of VGO are at 150 °C due to no values in higher temperature were available.
The density of VGO at 150 °C is estimated by multiplying the specific gravity of VGO at
150 °C and the density of water at 150 °C. The use of physical properties in a lower
temperature than the operating temperature will cause some inaccuracy in the
calculations.
Table 9. Physical properties of VGO.
Physical property Unit VGO
Specific gravity (150 °C) - 0.825*
Density (150 °C), ρ kg/m3 760
Kinematic viscosity (150 °C), μkin m2/s 2.60∙10-6**
Dynamic viscosity (150 °C), μdyn Pa∙s 2.32∙10-3
Source: *(Manning & Thompson, 1995), **(Jechura, 2015)
The density of 75 w% H3PO4 at 225 °C is calculated using Equation 18. The densities of
liquid pure water and H3PO4 at 225 °C were taken from Aspen Plus simulation program
and are 836 kg/m3 and 1700 kg/m3, respectively.
54
ρH3PO4,75% =mH3PO4
+ mH2O
mH3PO4
ρH3PO4
+mH2O
ρH2O
(18)
Therefore, the density of liquid, pressurized 75 w% H3PO4 at 225 °C is
ρH3PO4,75w% = 1351.5kg
m3≈ 1350
kg
m3
8.2 Equipment
The demetallization process consists of the static mixer and either of the decanter or the
electrostatic coalescer. In addition, H3PO4 is stored in a storage tank before pumping it
into the mixer. In this chapter, the preliminary design and sizing of the static mixer and
decanter are conducted. The electrostatic coalescer and the storage tank of H3PO4 are
discussed briefly in the end of this chapter.
8.2.1 Static mixer
In the static mixer, H3PO4 reacts with the metal impurities to form water soluble
compounds like metal phosphates. As the amount of H3PO4 is negligible compared to the
amount of VGO, the sizing of the static mixer is calculated using the values of VGO.
A Kenics mixer was chosen to be used as the static mixer in the demetallization process.
Kenics mixers are among the most commonly applied static mixers. Designed and
manufactured by Chemineer, the Kenics mixers consist of a variety of different mixer
designs. The pressure drop through the static mixer depends on the design and whether
the flow is laminar or turbulent. Moreover, the type of flow affects the design since some
mixers are especially designed for the laminar flow and some for turbulent. (Coker, 2007)
55
An injection mixer is a suitable design when one flow is much lower than the other
(Towler & Sinnott, 2013). As the flow rate of H3PO4 is significantly lower than the flow
rate of VGO, the streams are mixed using injection in the demetallization process. A
design of static injection mixer is can be seen in Figure 17.
Figure 17. Static injection mixer. Adapted from Towler & Sinnott (2013).
Dimensions of the mixer
The diameter of the static mixer is determined by calculating the economic pipe
diameter. The capital cost of a pipe increases with diameter, whereas the pumping costs
decrease with increasing diameter. A rule of thumb for the economic pipe diameter that
is used in oil refining is presented below (Equation 19). (Towler & Sinnott, 2013) In order
to obtain meters from the equation, the units inside the square brackets reduce the extra
units.
Di,optimum = 3.2 (m
ρ)
0.5
[(1 s
1 m)
0.5
] (19)
where Di = pipe inside diameter, m
m = mass flow rate, kg/s
56
ρ = density of fluid, kg/m3
Therefore,
Di,optimum = 1.06 m ≈ 1 m
The diameter of static mixer is rounded to 1 m. Reynolds number is calculated using
Equation 20 to determine whether the flow is laminar or turbulent (Menon, 2015).
Re =ρDu
μ (20)
where Re = Reynolds number
ρ = density of fluid, kg/m3
D = equivalent diameter, m
u = fluid velocity, m/s
μ = dynamic viscosity of fluid, Pa∙s
The fluid velocity u is calculated by dividing volumetric flow rate with the cross-sectional
area of the pipe. Therefore,
Re =Dm
μπ (D2)
2 = 45714
Below a Reynolds number of 2000, the flow in pipes will be laminar. The flow will be
critical when the Reynolds number is between 2000-4000 and turbulent above 4000.
Therefore, the fluid flow in the static mixer is turbulent. (Menon, 2015)
57
For the mixing of turbulent flow, a suitable option is to use a mixer which generates
controlled vortex structures. The mixing element geometry takes advantage of the
naturally occurring vortices induced by the element edges. A vortex generating Kenics
mixer called the HEV mixer is presented in Figure 18. (Coker, 2007)
Figure 18. Structure of HEV static mixer (Coker, 2007).
Although for the standard HEV mixer L/D ratio is 1, the L/D for the placing of the HEV
into the piping system is nominally 4 due to the vortexes which provide mixing
downstream of the mixer. Kenics has specified that three pipe diameters of straight pipe
need to be attached to the exit of the mixer. (Couper et al., 2012) To estimate the length
of the mixer, the amount of mixing elements need to be determined.
According to guidelines for sizing an AdmixerTM for liquid/liquid mixing, the number of
elements depends on the Reynolds number and the ratio of fluid viscosities and volumes.
In the demetallization process based on the guidelines the static mixer requires four
mixing elements because the Reynolds number is above 5000 and the volumetric ratio
between VGO and H3PO4 exceeds 100:1. (Admix, 1998) To assure the fluids are
completely mixed one additional mixing element is added in the preliminary design and
therefore, the total of five elements are utilized in the mixer. As the diameter of the
mixer is 1 m, the length of the mixing section is 20 m.
58
However, to provide sufficient reaction time the mixing section is followed by an empty
pipe without mixing elements. The liquid hourly space velocity (LHSV) is the reactant
liquid flow rate divided by reactor volume and is determined to be 3 h-1 in this
demetallization process. The length of the static mixer is calculated based on the LHSV
and the dimensions of the mixer are summarized in Table 10.
Table 10. Dimensions of the static mixer.
Equipment Volume
(m3) Cross-sectional
area (m2) Inside diameter
(m) Mixing section
length (m) Total length
(m)
Static mixer 132 0.785 1.0 20 168
The total length of the mixer is 168 m including the mixing section of 20 m. The mixer
requires a large space and therefore, it would need to be constructed further from the
HDS process as the space available close to the existing HDS process is limited. Due to
the great length, the mixer would rather have bends than be constructed of a straight
pipe. In the preliminary design, the mixer is constructed using straight pipes of 10 m
which are connected by 180° standard radius elbows to form a compact mixer design.
The ratio of elbow radius to pipe diameter is 1 for the standard radius elbow resulting in
1 m elbow radius and approximately 3 m elbow length. Therefore, the mixer consists of
13 straight pipes of 10 m length and 12 standard radius elbows of 180°. The amount and
type of elbows affect the pressure drop in the mixer which is discussed next in this
chapter.
Pressure drop
As the pressure drop is dependent on the design of the mixer, special data is required
from the manufacturer. Therefore, the pressure drop should be determined with the
assistance of the manufacturer. The pressure drop in the Kenics mixer can be determined
59
by multiplying the pressure drop in an empty pipe with a specific coefficient K. The
equation for calculating K in a turbulent flow is presented in Equation 21 and it must be
calculated using the data of the manufacturer. (Coker, 2007)
K = KOTB (21)
where KOT = factor related to the diameter and wall thickness of the mixer
B = factor related to Reynolds number
Some data for calculating the coefficient K has been published. Values for factor B can
be read from a graph published by Coker (2007). However, the values available for KOT
are only for mixers which inside diameters are in the range of about 0.15-0.3 meters. For
the data available, the values for KOT are between 21 and 41, mostly close to 30. (Coker,
2007) Therefore, an assumption that the KOT values would be within similar range for
larger mixers is made. To calculate the pressure drop in the mixer of demetallization
process, the factor KOT is estimated to be approximately 30. The manufacturer should be
contacted in order to get more accurate data. The factor B is read from the graph and
has a value of 1.95 (Coker, 2007). Therefore, the coefficient K is 58.5 which means the
pressure drop in the mixer is 58.5 times the pressure drop in an empty pipe.
The pressure drop in an empty pipe due to friction can be calculated using Equation 22.
The friction factor f is dependent on Reynolds number and relative roughness of the pipe.
The relative roughness is the result of dividing the absolute pipe roughness by the pipe
inside diameter. The absolute roughness of a commercial steel pipe is 0.046 mm and
therefore, the relative roughness is 4.6·10-5. As can be seen from Appendix 1, the friction
factor is 0.0026. (Towler & Sinnott, 2013)
ΔP = 8fL
Di
ρu2
2 (22)
60
where ΔP = pressure drop, Pa
f = friction factor
L = pipe length, m
Di = pipe inside diameter, m
ρ = density of fluid, kg/m3
u = fluid velocity, m/s
The pressure drop caused by the 180° elbows is estimated as a length of pipe that would
cause the same pressure loss. Additionally, the pressure drop caused by the inlet and
outlet of the mixer need to be taken into account when calculating the pressure drop in
the mixer. The pressure drop in the elbows and in the inlet and outlet of the mixer are
calculated using equivalent pipe diameters which are added to the length of the mixer in
Equation 22. The number of equivalent pipe diameters for a 90° standard radius elbow
is 30-40 and therefore, the number for a 180° standard radius elbow is estimated to be
approximately 80. The number of equivalent pipe diameters for the inlet is 50 and for
the outlet 25 for turbulent flow. (Towler & Sinnott, 2013)
The total length used in pressure drop calculation is 1203 m including the 20 m long
mixing section. The mixing section consists of 17 m of straight pipe and one 180° elbow
and therefore, the pressure drop in the 20 m long mixing section is calculated by
multiplying the pressure drop in a 97 m long empty pipe with the coefficient K. Instead
of adding the pressure drop caused by the inlet of the mixer to the mixing section, it is
included in the pressure drop of the empty pipe.
Therefore,
ΔPmixing section = 874 Pa
ΔPempty pipe = 170 Pa
61
As a result, the total pressure drop in the static mixer is 1044 Pa.
Materials
The selection of materials of construction has a significant impact on the operability and
economics of refining units. The 20 m long mixing section requires more resistant
material than the empty pipe following the mixing section in order to minimize the
corrosion caused by H3PO4. Although the concentration of phosphoric acid in the VGO
stream during the mixing is extremely small, corrosion may still occur.
Carbon steel is the most common construction material in refineries. Every effort is made
to use carbon steel because it is inexpensive, readily available and easily fabricated. It is
generally used in hydrotreating units that operate at temperatures below 260 °C.
(Garverick, 1994) Austenitic stainless steels and 20-type alloys such as 20Cb-3 are often
applied in equipment where H3PO4 is being handled (Davis, 2000). As the amount of acid
is relatively small, constructing the whole mixer using an expensive acid-resistant
material would not be cost-efficient. A suitable option for the mixer could be to use
austenitic stainless steel in the mixing section and carbon steel in the empty pipe. The
austenitic stainless steel chosen for the mixing section is alloy 904L due to its superior
corrosion resistant properties to phosphoric acid compared to conventional chrome
nickel stainless steels (Fine Tubes, 2011). The construction material affects the wall
thickness of the mixer which is calculated next in this chapter.
The static mixer needs to be well insulated to minimize heat losses and to keep the
desired temperature. The insulation can be placed around each pipe individually or
around a larger area including several pipes. The insulation needs to be easily removable
when maintenance is required.
62
Wall thickness
The American Society of Mechanical Engineers (ASME) has a standard for pipeline
transportation systems. The ASME Code for Pressure Piping is applied to determine the
wall thickness of the static mixer. The wall thickness without corrosion allowance is
calculated using Equation 23. (The American Society of Mechanical Engineers, 2012)
t =PDo
2SyEF (23)
where t = wall thickness without corrosion allowance, mm
P = internal design gauge pressure, kPa
Do = pipe outside diameter, mm
Sy = minimum yield strength of the pipe, kPa
E = weld joint factor
F = design factor based on nominal wall thickness
As only the inside diameter of the mixer is known, the outside diameter needs to be
replaced in the equation. The outside diameter Do equals the inside diameter Di summed
with two times the wall thickness t and two times the corrosion allowance c as can be
seen in Figure 19.
63
Figure 19. Cross-section of the mixer pipe; Do outside diameter, Di inside diameter, t wall thickness without corrosion allowance and c corrosion allowance.
Therefore, Equation 23 becomes
t =P(Di + 2t + 2c)
2SyEF=
P(Di + 2c)
2SyEF − 2P (24)
where Di = pipe inside diameter, mm
c = corrosion allowance, mm
Based on the ASME Code, the minimum yield strength Sy of 904L welded pipe is 220 MPa.
Carbon steel ASTM A53 grade A is used in the calculations of empty pipe and its minimum
yield strength is 207 MPa. The weld joint factor E is 1.00 and the value of design factor F
is 0.72 for both materials. (The American Society of Mechanical Engineers, 2012)
According to Hurme (2008), a typical corrosion allowance for stainless steel is 1 mm. For
carbon steel, a minimum corrosion allowance of 2.0 mm should be used when severe
corrosion is not expected and an allowance of 4.0 mm when more severe corrosion is
anticipated (Towler & Sinnott, 2013). Therefore, the mixing section has a corrosion
allowance of 1 mm and the empty carbon steel pipe an allowance of 2 mm.
64
A recommended margin between the normal operating pressure and the design pressure
is 10 % (Towler & Sinnott, 2013). As the operating pressure is 6.0 MPa in the mixer, the
design pressure will be 6.6 MPa and design gauge pressure 6.5 MPa. As a result, the wall
thicknesses t are
tmixing section = 21.4 mm ≈ 21 mm
tempty pipe = 22.9 mm ≈ 23 mm
The wall thicknesses with corrosion allowances are 22 mm for the mixing section and 25
mm for the empty pipe.
8.2.2 Decanter
The mixture of liquids leaving the mixer must be settled, coalesced and separated into
its liquid phases. A decanter is possible equipment used for separating the oil phase from
the aqueous phase. A horizontal cylindrical decanter is designed for the separation.
The aqueous phase consists of reaction products of the reaction between H3PO4 and
metal impurities. The density of aqueous phase is estimated by using the densities of
H3PO4 and pure solid metals. The densities and mass flow rates of the metal impurities
reacting with H3PO4 are shown in Table 11. Average density is calculated applying
Equation 18 and as a result, the average density of aqueous phase in the decanter is 1440
kg/m3. Laboratory experiments are required to determine the density more accurately.
65
Table 11. Densities and mass flow rates of components present in aqueous phase.
Substance Density (kg/m3)
Mass flow rate in aqueous phase (kg/h)
Ni 8900* 0.288
V 6065* 0.207
Fe 7870* 0.066
H3PO4, 75 w-% 1350 7.000
*Source: (Yaws, 2011)
Dimensions of the decanter
In a decanter the velocity of continuous phase, in this case the oil phase, should be less
than the settling velocity of the droplets of dispersed phase. The velocity of continuous
phase is calculated using Equation 25 and Stokes’ law is used to determine the settling
velocity of the droplets (Equation 26). (Thakore & Bhatt, 2007; Towler & Sinnott, 2013)
uc =Vc
Ai< ud (25)
where uc = velocity of the continuous phase, m/s
Vc = volumetric flow rate of continuous phase, m3/s
Ai = area of interface, m2
ud = settling velocity of the dispersed phase droplets, m/s
ud =dd
2g(ρd − ρc)
18μc (26)
where dd = droplet diameter, m
g = acceleration of gravity = 9.81 m/s2
66
ρd = density of dispersed phase, kg/m3
ρc = density of continuous phase, kg/m3
μc = viscosity of continuous phase, Pa∙s
The droplet size of 150 μm is often used for the sizing of a decanter because it is well
below the droplet sizes normally found in decanter feeds. In a horizontal cylindrical
decanter, the interfacial area Ai depends on the position of the interface and is calculated
using Equation 27. Equation 28 is used to calculate the width of the interface needed to
determine the interfacial area. The geometry of the horizontal decanter is presented in
Figure 20. (Thakore & Bhatt, 2007; Towler & Sinnott, 2013)
Ai = wl (27)
w = 2√2rz − z2 (28)
where w = width of interface, m
l = length of decanter, m
r = radius of decanter, m
z = height of interface from the base of the vessel, m
67
Figure 20. Geometry of horizontal decanter: r = radius of decanter, w = width of interface and z = height of the interface from the base of the vessel (Towler & Sinnott, 2013).
First, the settling velocity ud is calculated using Equation 26 and subsequently, the
minimum interfacial area Ai is obtained from Equation 25. The volumetric flow rate of
the oil phase used in Equation 25 is the VGO feed flow rate subtracted by the flow rate
of removed metal impurities.
ud = 3.58 mm/s
Ai,min =Vc
ud=
mc
ρcud= 30.64 m2
As 7 kg/h of H3PO4 is mixed with 300 t/h of VGO, the amount of aqueous phase is
extremely small compared to the oil phase. Due to the minuscule amount of aqueous
phase, the position of the interface will be below the centre line of the decanter. The
height of the liquid interface needs to be controlled accurately using a level instrument
because of the small quantity of aqueous phase (Towler & Sinnott, 2013). The level
instrument can control the height of the interphase by controlling the flow rate of
aqueous phase leaving the decanter. In the design of the demetallization process, the
position of the interphase is fixed at a height of Di/4. Equation 28 becomes
68
w = 2√DD
4− (
D
4)
2
=D
2√3
For a horizontal decanter, the length is approximated to be four times the diameter
(Thakore & Bhatt, 2007). Therefore, the minimum diameter of the decanter can be
calculated using Equation 27.
Ai,min =D
2√3 ∙ 4D = √12Dmin
2 → Dmin = 2.97 m
The minimum diameter required is 3.0 m and the length is 11.9 m. The large diameter
causes the wall thickness of the decanter also to be large. The wall thickness is discussed
later in this chapter.
The take-off locations can be determined by making a pressure balance. Neglecting the
friction loss in the pipes, the pressure exerted by the combined height of the heavy and
light liquid in the decanter must be balanced by the height of the heavy liquid in the take-
off leg as can be seen in Figure 21. The pressure balance is presented in Equation 29.
(Thakore & Bhatt, 2007; Towler & Sinnott, 2013)
69
Figure 21. Location of light and heavy liquid take-offs; zi height of the interface, zl height of light liquid take-off and zh height of heavy liquid take-off (Thakore & Bhatt, 2007).
(zl−zi)ρlg + ziρhg = zhρhg (29)
where ρl = density of the light phase, kg/m3
ρh = density of the heavy phase, kg/m3
g = gravitational acceleration = 9.81 m/s2
For a horizontal decanter, a good take-off height for the light phase is around 0.8 times
the diameter (Thakore & Bhatt, 2007). As the height of the interface is one fourth of the
diameter, the height of the heavy phase take-off can be calculated from the pressure
balance.
zi = 0.74 m
zl = 2.38 m
zh = 1.61 m
70
In order to minimize entrainment by the jet of liquid entering the decanter, the inlet
velocity should be kept below 1 m/s. The inlet pipe diameter can be calculated using
Equation 30. (Towler & Sinnott, 2013)
Dinlet pipe = √4A
π=
√4 (
mh
ρh+
ml
ρl)
πu (30)
where A = cross-sectional area of inlet pipe, m2
mh = mass flow rate of heavy phase, kg/s
ml = mass flow rate of light phase, kg/s
u = inlet velocity, m/s
Therefore,
Dinlet pipe = 0.374 m ≈ 0.4 m
The inlet pipe diameter should be approximately 0.4 m in order to keep the inlet velocity
below 1 m/s.
Materials
Carbon steel is the preferred construction material because it is inexpensive and readily
available. As phosphoric acid reacts with the metal impurities in the mixer, corrosion due
to H3PO4 does not occur in the decanter enabling the use of carbon steel. Therefore, the
decanter is constructed of carbon steel without any protective linings.
71
Wall thickness
For design purposes, the pressure vessels are divided into two classes depending on the
ratio of the wall thickness to vessel diameter. If the ratio is less than 1:10, the vessel is
thin-walled and if it is above that value, the vessel is considered to be thick-walled. The
decanter has a large diameter and therefore, the ratio will be less than 1:10 so it will be
classified as a thin-walled vessel. (Towler & Sinnott, 2013)
The principal stresses affecting the wall of a vessel due to a pressure load are shown in
Figure 22. As the wall is thin, the radial stress σ3 will be small and can be neglected in
comparison with the other stresses. The longitudinal stress σ1 and the circumferential
stress σ2 can be taken as constant over the wall thickness. The majority of the vessels
used in chemical industries are classified as thin-walled vessels. (Towler & Sinnott, 2013)
Figure 22. Principal stresses in a pressure vessel wall: σ1 longitudinal stress, σ2
circumferential stress and σ3 radial stress (Towler & Sinnott, 2013).
The ASME Boiler & Pressure Vessel Code (BPVC) regulates the design and construction
of boilers and pressure vessels. The minimum wall thicknesses due to longitudinal and
circumferential stresses are calculated using Equations 31 and 32, respectively. The
ASME BPVC specifies that the minimum wall thickness of a pressure vessel is the greater
72
value determined from these equations. (The American Society of Mechanical Engineers,
2010)
t =PiDi
4SE + 0.8Pi (31)
t =PiDi
2SE − 1.2Pi (32)
where t = vessel wall thickness, mm
Pi = internal pressure, kPa
Di = inside diameter of vessel, mm
S = maximum allowable stress, kPa
E = joint efficiency
The operating pressure in decanter is close to the operating pressure in mixer because
of the relatively small pressure drop in the mixer. Therefore, 6.6 MPa is also used as the
design pressure for the decanter. Maximum allowable stresses for different materials
are presented in ASME BPVC. Carbon steel SA/EN 10028-2 grade P355GH is used for the
design of the decanter and the maximum allowable stress for that carbon steel is 137
MPa (The American Society of Mechanical Engineers, 2015). A good value for the joint
efficiency E is 0.7 for preliminary design (Escoe, 2008).
Therefore, the minimum wall thickness due to longitudinal stress is
t = 51 mm
and the minimum thickness due to circumferential stress is
73
t = 108 mm
As a result, the minimum wall thickness of the decanter would be 108 mm. However,
that great thickness implies that the mass of the decanter vessel will also be great.
Applying two smaller decanters of 2.1 m diameter each in parallel would reduce the wall
thickness to 75 mm. Similarly, using three decanters of 1.7 m diameter in parallel would
further reduce the wall thickness to 61 mm. All these decanter configurations provide
the same interfacial area for the separation. Applying several decanters reduces the wall
thickness and the weight of the vessels but they require more land area.
A corrosion allowance is not included in the wall thicknesses calculated using Equations
31 and 32. The corrosion allowance should be based on experience with the material
under similar conditions than the operating ones. The equal corrosion allowance of 2.0
mm is used in the decanter as in the empty carbon steel pipe of the mixer. The
dimensions of different decanter configurations are summarized in Table 12.
Table 12. Dimensions of different decanter configurations.
Number of decanters
Diameter (m)
Length (m)
Wall thickness with 2 mm corrosion allowance (mm)
1 3.0 11.9 110
2 2.1 8.4 77
3 1.7 6.9 63
Decanter heads
Torispherical, ellipsoidal and hemispherical heads are commonly used for pressure
vessels. Different head types are presented in Figure 23. The torispherical heads are
usually used for pressures up to 20 bar, the ellipsoidal heads for pressures between 20
74
and 100 bar and the hemispherical heads for pressures above 100 bar. As the design
pressure is 66 bar, the decanter is designed with ellipsoidal heads. (Hall, 2012)
Figure 23. Pressure vessel heads: (a) hemispherical, (b) ellipsoidal and (c) torispherical (Towler & Sinnott, 2013).
The most standard ellipsoidal heads have a major-to-minor axis ratio of 2:1 (Towler &
Sinnott, 2013). The depth of dish a, which can be seen in Figure 24, is the distance from
a tangent line to the end of head. The tangent line is located in the beginning of the
curvature of the head. Because the head includes a short straight section, the flange, the
tangent-to-tangent distance of the vessel is greater than distance between the weld
seams at the heads. The straight flange is roughly from 10 cm to 30 cm depending on
diameter and thickness. The depth of dish a for an ellipsoidal vessel is one fourth of the
inside diameter of the vessel. (Hall, 2012) The depth of dishes is presented in Table 13.
75
Figure 24. Depth of an ellipsoidal head; a is the distance from the tangent line to the end of head and D is the inside diameter of the vessel (Hall, 2012).
The wall thickness of the ellipsoidal head is calculated using Equation 33 (The American
Society of Mechanical Engineers, 2010).
t =PiDi
2SE − 0.2Pi (33)
where Pi = internal pressure, kPa
Di = inside diameter of vessel, mm
S = maximum allowable stress, kPa
E = joint efficiency
If formed by pressing, the head has no joints and the joint efficiency E will be 1 (Towler
& Sinnott, 2013). The head thicknesses of different decanter configurations are
presented in Table 13.
76
Table 13. Depth of dishes and head thicknesses of different decanter configurations.
Number of decanters
Inside diameter (m)
Depth of dish (m)
Head thickness using Equation 33 (mm)
1 3.0 0.75 70
2 2.1 0.53 51
3 1.7 0.43 41
The head thickness of a decanter is rounded to the value of the shell wall thickness
(Towler & Sinnott, 2013).
Decanter support
The decanter vessel must be supported. Generally, horizontal tanks and pressure vessels
are supported on two vertical cradles called saddles. The use of more than two saddles
is unnecessary and should be avoided as two saddles provide a high tolerance for soil
settlement with no change in shell stresses. Usually, one end of the vessel is anchored
and the other end sliding to enable the thermal expansion of the vessel. The saddle can
be constructed of concrete or steel and it consists of various parts: the web, base plate,
ribs and wear plate. (Moss & Basic, 2013)
The contact angle of vessel and wear plate should not be less than 120° and usually, it is
between 120-150°. Typical saddle design dimensions for vessels with different diameters
are shown in Figure 25 and Table 14. (Towler & Sinnott, 2013) The saddle designs listed
in Table 14 present the closest values available to the decanter diameters of 3.0 m, 2.1
m and 1.7 m.
77
Figure 25. Dimensions of a saddle support: V distance from the bottom of the base plate to the center line of the vessel, J ribs spacing, C base plate length, Y base plate width, t1 rib and web thickness, t2 base plate thickness (Towler & Sinnott, 2013).
Table 14. Typical values for the saddle support design (Towler & Sinnott, 2013).
Dimensions (m) Dimensions (mm)
Vessel diameter (m)
V J C Y t1 t2
1.6 0.98 0.350 1.41 0.20 10 12
1.8 1.08 0.405 1.59 0.20 10 12
2.0 1.18 0.450 1.77 0.20 10 12
2.2 1.28 0.520 1.95 0.23 12 16
3.0 1.68 0.665 2.64 0.25 12 16
8.2.3 Electrostatic coalescer
Alternative equipment for the separation of the phases is an electrostatic coalescer. The
droplets of the aqueous phase may be so fine in diameter that they do not settle by
gravity (Fahim et al., 2010). Moreover, the small amount of aqueous phase compared to
the amount of oil phase can complicate the coalescing of the aqueous phase in the
decanter.
78
Coalescing of the small water drops in oil can be done by using a high-voltage electrical
field. When a non-conductive liquid containing dispersed conductive liquid is subjected
to an electrical field, the conductive droplets are caused to combine. The electric field
ionizes the water droplets which coalesce to produce larger drops that can be settled by
gravity. Electrostatic coalescers are used for desalting crude oils. (Fahim et al., 2010)
A horizontal electrostatic coalescer is presented in Figure 26. The coalescer can contain
two or more electrodes. One electrode is grounded to the vessel and the other
suspended by insulators. An alternating current is used and the usual applied voltage
ranges from 10 to 35 kV. The electrical charges of the electrodes reverse many times a
second causing the water droplets to be in a rapid back-and-forth motion. The possibility
of water droplets colliding with each other increases with greater motion of the droplets.
The intensity of the electrostatic field is controlled by the applied voltage and the spacing
of the electrodes. (Arnold & Stewart, 2008)
79
Figure 26. Cutaway of a horizontal electrostatic coalescer (Arnold & Stewart, 2008).
In the design of coalescer, preventing the water level from reaching the height of the
electrodes is important. The salts present in the aqueous phase make it a very good
conductor of electric currents. Therefore, a contact with the electrodes may short out
the electrode grid or the transformer. The sizing of electrode grid area requires
laboratory testing because the coalescence of droplets in an electric field is dependent
on the characteristics of the specific mixture being treated. (Arnold & Stewart, 2008)
Coalescer vessels are sized for a certain volume flow per square meter of grid area.
According to Arnold & Stewart (2008), procedures for designing electrostatic grids have
not been published. As the coalescence of water droplets in an electric field is extremely
dependent on the characteristics of the treated emulsion, it is unlikely that a general
relationship of water droplet size to use in the settling equations can be developed.
80
8.2.4 Storage tank for phosphoric acid
Phosphoric acid requires a storage tank from where it is fed to the static mixer. Raw
material storage sizing depends much on the logistics and economical delivery lots.
Transporting H3PO4 in lots that are smaller than one full truck lot is relatively more
expensive than transporting full truck lots. Therefore, the sizing of the storage tank is
based on the capacity of a normal truck lot which is between 10 and 20 m3 (American
Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., 2012). Because the
mass flow rate of H3PO4 is only 7 kg/h, a truck with 10 m3 capacity can be used. To have
some safety margin if the delivery is late or the consumption of the acid is greater than
estimated, a two-week acid consumption with 7 kg/h flow rate is added to the volume
of storage tank. Depending on the temperature of the acid, the total volume flow rate
in 14 days is 1.5-1.7 m3 and as a result, the storage tank would be 11.7 m3. However, the
tank should not be more than 85-90 % full (Hurme, 2008). Therefore, a tank of 14 m3
would be sufficient.
The storage tank for H3PO4 requires a corrosion resistant material. The tank can be
constructed of the same austenitic stainless steel that is used for the mixing section. The
alloy 904L has good corrosion resistance to phosphoric acid (Fine Tubes, 2011).
A containment dike is applied to prevent major contamination to surrounding areas in
case of leaks and spills. The dike must hold the content of the storage tank if a leak
occurs. (Parisher & Rhea, 2012) The storage tanks are usually placed in a common area
of the facility known as a tank farm. The tank farm should be placed between the loading
and unloading facilities and the process units they serve. The area must be easily
accessible in order to organize the unloading of the tanks practically and safely. (Towler
& Sinnott, 2013)
81
8.3 Profitability
The costs related to the demetallization process are divided into investment and
operating costs. The investment cost includes, for example, the costs of purchasing and
installing the equipment while the operating costs result from purchasing the raw
materials and changing the catalyst. In the end of this chapter, a sensitivity analysis is
conducted and the profitability of the demetallization process is discussed.
8.3.1 Investment cost
The purchased equipment costs can be estimated in various ways. The best source of
purchase costs is recent data on actual prices paid for similar equipment. Reliable
information on equipment costs can also be found in professional cost engineering
literature. In addition, different cost correlations and cost estimating software, for
example Aspen Technology’s Aspen ICARUSTM Technology, can be used for estimating
chemical plant costs. (Towler & Sinnott, 2013)
Static mixer
The purchase costs of the mixing section and empty pipe are estimated separately. The
sum of these costs forms the purchase cost of the static mixer.
ZAIN Technologies has published typical market prices of static mixers with diameters
from 2 to 12 inches. The mixers are constructed from 316L stainless steel and they
contain 3 or 6 helical mixing elements. (ZAIN Technologies, 2016) The prices are assumed
to be updated to present values as they are taken from the company website. The price
of a mixer, the diameter of which is 1 m, is roughly estimated by extrapolating the prices
of 6-element mixers. The purchase cost of the mixing section in the demetallization
process is estimated to be 5 times the price obtained by extrapolating as the mixing
82
section consists of 5 HEV mixing elements. As a result, the purchase cost of a mixer made
of 316L stainless steel is US$125,000.
To estimate the purchase cost of the mixer constructed from 904L, a material cost factor
is required. According to Steel Tank Institute/Steel Plate Fabricators Association, relative
cost ratio between 904L and 316L is 2.16 (Steel Tank Institute/Steel Plate Fabricators
Association, 2012). Therefore, the purchase cost of the mixing section made of 904L is
US$270,000.
Finally, the U.S. dollars are converted into euros by applying the currency rate of U.S.
dollar on 17th of February 2016. On that date, one euro was worth 1.11332 U.S. dollars
(Kauppalehti, 2016). Therefore, 270 k$ equals 243 k€.
The cost of the empty pipe is estimated based on its weight. The weight of the empty
pipe is calculated by multiplying the volume of carbon steel in the empty pipe by its
density. The density of carbon steel used in the calculations is 7800 kg/m3. The world
carbon steel price was taken from the tables of Management Engineering & Production
Services. The price in 2015 was approximately 500 US$/ton (Management Engineering &
Production Services, 2016). The previously used currency rate of U.S. dollar is applied to
convert the cost in euros. The purchase cost of the carbon steel pipe is estimated to be
twice the material cost. The cost of the empty carbon steel pipe is presented in Table 15.
Table 15. Cost of the empty carbon steel pipe required for the static mixer.
Inside diameter
(m) Length
(m)
Wall thickness
(mm) Weight
(kg)
Cost of carbon
steel (k$)
Cost of carbon
steel (k€)
Purchase cost of
pipe (k€)
Carbon steel pipe 1 148 25 92930 46.5 41.7 83.5
83
The total purchase cost of the static mixer is the sum of the costs of mixing section and
the empty carbon steel pipe. As can be seen in Table 16, the purchase cost of static mixer
is 326.5 k€. The installed cost of the mixer is preliminarily estimated to be three times
the purchase cost. Therefore, the cost of installed static mixer is 980 k€.
Table 16. Purchase cost of static mixer.
Equipment Purchase cost (k€)
Mixing section 243
Empty carbon steel pipe 83.5
Static mixer 326.5
Decanter
The purchase cost of a horizontal carbon steel pressure vessel can be calculated using
Equation 34 which is based on the weight of the shell and two 2:1 ellipsoidal heads. The
purchase cost is in year 2000 basis and includes an allowance for platform, ladders and
manholes. (Seider et al., 2004)
Cp = exp(8.717 − 0.2330 ln W + 0.04333(ln W)2) + 1580Di0.20294 (34)
where Cp = purchase cost of horizontal carbon steel pressure vessel, $
W = weight of shell and two heads, lb
Di = inside diameter of vessel, ft
The costs are calculated for three different decanter configurations: one decanter of 3.0
m diameter, two decanters of 2.1 m diameter each and three decanters of 1.7 m
diameter each. The weight of vessel shells is presented in Table 17.
84
Table 17. Shell weight of decanters.
Number of decanters
Diameter (m)
Length (m)
Wall thickness
(mm)
Volume of carbon steel in one
decanter (m3) Shell weight of
one decanter (kg)
1 3.0 11.9 110 12.7 98800
2 2.1 8.4 77 4.4 34600
3 1.7 6.9 63 2.4 18900
The weight of a vessel head can be accurately determined by calculating the volume of
a circular blank. The blank diameter of a 2:1 ellipsoidal head is calculated using Equation
35. (Escoe, 2008)
Db = 1.22Di + 2(S. F. ) + t (35)
where Db = blank diameter, m
Di = inside diameter, m
S.F. = straight flange of head, m
t = head thickness, m
For the straight flange a value of 0.2 m is used in the preliminary design. The weight of
the vessel head is obtained by multiplying the volume of the circular blank with the
density of carbon steel. The results are shown in Table 18.
85
Table 18. Vessel head weights and the total weight of shells and heads.
Number of decanters
Inside diameter
(m)
Straight flange of head (m)
Head thickness
(mm)
Blank diameter
(m)
Weight of one head
(kg)
1 3.0 0.2 110 4.14 11530
2 2.1 0.2 77 3.04 4360
3 1.7 0.2 63 2.56 2520
To calculate the purchase cost of the decanters, the diameters need to be converted into
feet and the weights into pounds. The purchase costs of different decanter
configurations are shown in Table 19.
Table 19. Total purchase cost of decanters in U.S. dollars in 2000.
Number of
decanters Di(m) Di(ft)
Weight of one shell and two
heads (kg)
Weight of one shell and two
heads (lb)
Purchase cost of one decanter in
2000 ($)
Total purchase
cost in 2000 ($)
1 3.0 9.8 121890 268720 292000 292000
2 2.1 6.9 43300 95450 128100 256200
3 1.7 5.6 23900 52680 83500 250500
The costs calculated using Equation 34 are in year 2000 basis. To update the costs to
present value a chemical engineering plant cost index (CEPCI) is applied. The CEPCI in
2000 was 394. (Seider et al., 2004) The most recent CEPCI available is for April 2015 when
the value was 562.9 (Chemical Engineering, 2015). Therefore, the costs in April 2015 can
be calculated using Equation 36 (Towler & Sinnott, 2013).
CApril 2015 = C2000
CEPCIApril 2015
CEPCI2000 (36)
86
The purchase costs in April 2015 can be seen in Table 20. The currency is converted into
euros using the same currency rate as previously. The cost of an installed carbon steel
pressure vessel can be estimated by multiplying the purchase cost by a factor of 4.1. The
installed cost includes the installation of equipment, cost of site preparations,
foundations, structures, piping, engineering, fee of contractor and supervision. (Hurme,
2008)
Table 20. Total purchase and installed cost of decanters in 2015.
Number of decanters
Total purchase cost in 2015 (k$)
Total purchase cost in 2015 (k€)
Total installed cost of decanters in 2015 (k€)
1 417 375 1537
2 366 329 1348
3 358 322 1318
Applying three smaller decanters has the lowest purchase cost. However, the difference
in purchase costs between two decanters and three decanters is not great. Even though
the decanters are smaller, applying three decanters will require more space than one
larger decanter. Furthermore, more piping is needed as the VGO stream is divided into
three decanters and after combined back together before feeding it to the reactor. The
large decanter weights more than the other decanter configurations and therefore, it
requires stronger and more durable foundations that can withstand the combined
weight of the decanter and the VGO stream.
Due to the lowest purchase cost, three decanters are used in the sensitivity analysis
which is conducted in chapter 8.3.3. In addition, if three decanters are applied in parallel,
the process would not need to be shut down if problems occurred in one of the
decanters. By reducing the flow rate, one decanter could be under maintenance while
the other two remain working.
87
8.3.2 Operating costs
The operating costs of demetallization process are discussed in this chapter. The
operating costs consist of the purchase costs of raw materials and the cost of changing
the catalysts.
Purchase cost of VGO and phosphoric acid
If a VGO feed with 10 ppm of metal impurities could be used, the purchase cost would
be lower compared to the cleaner VGO feed currently utilized in the refinery. Currently,
VGO with high impurity levels is approximately 12 US$/ton cheaper than the cleaner
VGO. However, the same VGO feed is used in both cases 1 and 2 so the purchase cost of
VGO will not affect the comparison of these two cases.
Based on a market analysis report of PotashCorp, the price of phosphoric acid was 600-
1100 US$/ton between January 2010 and January 2014. The average price was
approximately 800 US$/ton. (PotashCorp, 2014)
Cost of changing the catalyst
The amount of metal impurities accumulating on the guard bed catalyst depends on the
phase separation in the decanter or electrostatic coalescer. As the amount of the
aqueous phase is extremely small compared to the oil phase, the aqueous phase might
be so dispersed that the coalescence and separation into two phases will not occur
completely. As a result, a part of the aqueous phase flows into the HDS reactors with the
oil phase.
The catalyst lifetime is calculated assuming that the phases are separated perfectly.
However, depending on the efficiency of the separation method the catalyst lifetime may
be slightly shorter. The most of the metal impurities are present in the oil phase despite
the use of phosphoric acid and therefore, the small possible part of aqueous phase
88
entering the reactors will not significantly reduce the catalyst lifetime. The amount of
accumulated metal impurities in the guard bed catalyst is presented in Table 21.
Table 21. Accumulation of metal impurities in guard bed after demetallization.
Accumulation of impurities in the guard bed (kg)
Week
Feed stream
(t/h)
Feed stream (t/wk) As Ca Fe Na V Ni Total
1 300 50400 1.3 11.3 9.8 25.2 29.1 25.7 102
2 300 50400 2.5 22.7 19.7 50.4 58.2 51.4 205
3 300 50400 3.8 34.0 29.5 75.6 87.3 77.1 307
4 300 50400 5.0 45.4 39.3 100.8 116.4 102.8 410
5 300 50400 6.3 56.7 49.1 126.0 145.5 128.5 512
6 300 50400 7.6 68.0 59.0 151.2 174.6 154.2 615
7 300 50400 8.8 79.4 68.8 176.4 203.7 179.9 717
8 300 50400 10.1 90.7 78.6 201.6 232.8 205.6 820
9 300 50400 11.3 102.1 88.5 226.8 262.0 231.3 922
10 300 50400 12.6 113.4 98.3 252.0 291.1 257.0 1024
11 300 50400 13.9 124.7 108.1 277.2 320.2 282.7 1127
12 300 50400 15.1 136.1 117.9 302.4 349.3 308.4 1229
As discussed earlier, the adsorption capacity of the guard bed is 1.2 tons of metal
impurities. The capacity is full during the week 12. Using the research data of Neste, the
HDS catalyst will be deactivated already during week 10 and therefore, the catalyst
lifetime is 9 weeks. Applying the demetallization process extends the catalyst lifetime
only by one week. The reason why the catalyst lifetime does not extend more is that
phosphoric acid reduces only the concentration of nickel, vanadium and iron. Moreover,
the removal rates of nickel, vanadium and iron are fairly low. The rest of the impurities
continue to deactivate the catalyst as their amount is not affected by phosphoric acid.
The costs of a process shutdown in Neste cannot be published in the thesis. Therefore,
the cost of changing the catalysts during one process shutdown is calculated using costs
published by Gorra et al. (1993). According to their article, one shutdown lasted 10 days
and the lost profit was 39 US$/ton. The average catalyst handling cost was US$61,000
89
and the catalyst price 5 US$/kg. (Gorra et al., 1993) The labour costs were not published
but to the best of knowledge, the labour costs form only a small part of the total costs.
Especially compared to the lost profit, the labour costs are relatively low.
Costs of changing the catalyst during one shutdown are presented in Table 22. The costs
in 1993 are converted into current values by taking into account the inflation. The mean
value of inflation in the United States between 1993 and 2016 is approximately 2.2 %
(Trading Economics, 2016). The previously used currency rate of U.S. dollar is applied for
the conversion into euros.
Table 22. Costs of changing the catalysts during one process shutdown.
Costs of changing the catalyst
In 1993 (k$)
In 2016 (k$)
In 2016 (k€)
Lost production 2808 4632 4161
Catalyst handling cost 61 101 90
Catalyst price 1000 1650 1482
Total cost 3869 6383 5733
Based on the article by Gorra et al. (1993), the cost of one process shutdown is 5.7 M€.
The cost of changing the catalyst during one shutdown is the same for case 1. Compared
to the investment costs, the costs caused by the shutdowns are significantly greater. A
sensitivity analysis is conducted in the following chapter and the impacts of different
parameters on the profitability are discussed.
8.3.3 Sensitivity analysis
A sensitivity analysis is a tool for examining the effects of uncertainties in the forecasts
on the viability of a project. The purpose is to identify the parameters that have a
significant impact on the project viability and investigate the effect by varying each
parameter individually. The result of the sensitivity analysis is often presented as a plot
90
of net present value (NPV) versus the parameter studied. The NPV is strongly dependent
on the interest rate and the time period studied. (Towler & Sinnott, 2013) For the
demetallization process, the NPV is calculated for a 10-year operating period with a 15
% interest rate. The parameters studied in the sensitivity analysis are the number of
shutdowns, investment cost and purchase cost difference between low-quality and
cleaner VGO.
For the phase separation, three decanters are applied. Therefore, the investment costs
used in the sensitivity analysis consist of the installed costs of static mixer and three
decanters. The investment cost is 2298 k€.
As the HDS catalyst lifetime is 9 weeks and each shutdown lasts 10 days, 5 catalyst
change shutdowns are needed during each year. As a result, the process is under
operation 45 weeks per year. The possibility to utilize low-quality VGO with high impurity
levels saves US$12 per ton in VGO purchase costs. The annual savings due to cheaper
VGO are shown in Table 23. The savings are equal each year during the 10-year period.
Table 23. Annual savings due to the use of cheaper VGO.
Period Number of
operating weeks Processed VGO (tons)
Savings due to cheaper VGO (k$)
Savings due to cheaper VGO (k€)
1 year 45 2268000 27216 24446
The annual costs of purchasing H3PO4 and changing the catalyst are presented in Table
24. In the calculations, 800 US$/ton is used as the purchase price for the acid. The total
annual cash flow is calculated by subtracting the H3PO4 purchase costs and catalyst
changing costs from the savings shown in Table 23. The total cash flow of each year is
then discounted in order to obtain the present value. The present values of the total cash
flows are presented in Table 25 and in Figure 27. The investment in the equipment occurs
in year 0.
91
Table 24. Annual cost of purchasing H3PO4 and changing the catalysts.
Period
Number of operating
weeks Number of shutdowns
Amount of H3PO4 used
(tons)
Cost of purchasing H3PO4 (k$)
Cost of purchasing H3PO4 (k€)
Cost of changing the catalysts (k€)
1 year 45 5 53 42 38 28665
Table 25. Present value of total cash flow.
Year Discounting factor when
15 % interest rate Total cash flow
(k€) Present value of
total cash flow (k€)
0 1.0000 -2298 -2298
1 0.8696 -4257 -3702
2 0.7561 -4257 -3219
3 0.6575 -4257 -2799
4 0.5718 -4257 -2434
5 0.4972 -4257 -2117
6 0.4323 -4257 -1840
7 0.3759 -4257 -1600
8 0.3269 -4257 -1392
9 0.2843 -4257 -1210
10 0.2472 -4257 -1052
Figure 27. Present value of total annual cash flow.
-4000
-3500
-3000
-2500
-2000
-1500
-1000
-500
0
0 1 2 3 4 5 6 7 8 9 10
Pre
sen
t va
lue
(k€
)
Year
Present value of total annual cash flow
92
The NPV is the sum of the present values of the cash flow and therefore, the NPV of the
10-year period in the demetallization process is -23664 k€. As the NPV is negative,
investing in the demetallization process is not a cost-effective option with the current
prices and costs.
The process viability depends on the number of shutdowns, investment cost and
purchase cost difference between low-quality and cleaner VGO. To study the effects of
these parameters on the process viability, each of the parameters are individually varied
in the range from -50 % to +50 %. The results are presented as a plot of NPV versus
change in parameter value and can be seen in Figure 28.
Figure 28. Sensitivity analysis.
-100
-80
-60
-40
-20
0
20
40
60
-50 -25 0 25 50
NP
V (
M€
)
Change in parameter value (%)
Sensitivity analysis
Number of shutdowns
Purchase cost differencebetween VGO feeds
Investment cost
93
From Figure 28 can be seen that the number of shutdowns and the difference in VGO
purchase costs have the strongest impact on the viability of the process. The changes in
investment costs do not significantly affect the NPV. The NPV becomes positive if the
number of shutdowns is reduced at least by 17 % or the difference in VGO purchase costs
increases by 20 % or more. These values are summarized in Table 26. In order to reduce
the number of shutdowns by 17 %, the nickel, vanadium and iron levels must be reduced
by additional 45 %. As a result, the required removal rates for nickel, vanadium and iron
are 63 w%, 58 w% and 57 w%, respectively.
Table 26. Changes in parameter value that convert the NPV positive.
Parameter Current
value Value which converts
NPV positive Change compared to current value
Number of shutdowns during 1 year 5 4.15 -17 %
Purchase cost difference between low-quality and cleaner VGO ($/ton) 12 14.4 +20 %
8.4 Safety
Safety issues form an important part of process design. If the designed process is not
safe to operate, it is not viable. All safety risks must be eliminated or minimized to an
acceptable level.
Overpressure is one of the most serious hazards in a chemical plant. Overpressure will
occur when mass, moles or energy accumulate in a contained space with a restricted
outflow. For example blocked outlet, valve failure or external fire can be a cause of
overpressure in the demetallization process. To ensure that the pressure inside a vessel
cannot rise to an unsafe level, pressure relief devices must be installed. Pressure relief
devices are an essential requirement for the safe use of pressure vessels as they provide
94
a safe means of relieving overpressure. (Towler & Sinnott, 2013) Therefore, each
decanter must contain a pressure relief device.
Generally, high temperatures can arise from the loss of control of reactors and heaters
and from external fires (Towler & Sinnott, 2013). As the amount of metal impurities in
the VGO is extremely small, the heat of reaction will not significantly affect the
temperature of the VGO stream. Therefore, the more probable cause of high
temperature in the demetallization process is an external fire. The safety of the whole
chemical plant and the design of the layout contribute to minimizing the probability of
external fires. Secure water supplies for firefighting must be available and at least two
escape routes provided from each level in process buildings (Towler & Sinnott, 2013).
In addition to high temperatures, also too low temperatures cause problems in operating
the process. Low temperatures will increase the viscosity of VGO and therefore, restrict
its flow through the pipes and equipment. If problems occur in maintaining the
temperature of the VGO feed storage tank in the right level, the VGO becomes too
viscous to flow causing blockages.
The reactions occurring in the mixer need to be studied in order to find the optimum
reaction conditions and to identify side reactions. Reaction products can be hazardous
and require special handling which makes their identification important.
The probability of leaks should be minimized by careful equipment design. The corrosion
can be minimized by selecting suitable materials and operating conditions. The storage
tank of H3PO4, the pipeline feeding the acid to the mixer and the mixing section of the
static mixer have the most corrosive condition. The flow rate of H3PO4 into the mixer
must be controlled in relation to the VGO stream. If feeding of VGO is stopped due to
problems in the process, a flow rate controller must stop the acid feed in order to not
waste acid and to avoid corrosion problems in the mixer and decanters. Adequate
maintenance procedures secure that the equipment stay in good condition during
95
operation. In addition, the plant needs to be supplied with drainage and sewer systems
which collect the possible runoff, firefighting water and rainwater for waste treatment
(Towler & Sinnott, 2013).
The sparking of electrical equipment, such as motors, is a potential source of ignition
(Towler & Sinnott, 2013). As the mixing is obtained without a moving agitator, the static
mixer has no motor. Therefore, the process contains one ignition source less. All
electrical equipment must be grounded adequately (Towler & Sinnott, 2013).
If three decanters were applied in parallel, the process would not need to be shut down
if problems occurred in one of the decanters. By reducing the flow rate, one decanter
could be under maintenance while the other two are working. The static mixer does not
have a duplicate system. Because the mixer is a pipe which does not contain moving
parts, it requires only little maintenance and therefore, applying a duplicate system is
unnecessary. The maintenance required due to fouling can be performed during planned
process shutdowns. If sudden problems in operating the process occur and require a
process shutdown, there must be secure means of controlling the process into a stable
and safe state.
The aqueous phase separated in the decanter or electrostatic coalescer must be treated.
The amount of aqueous phase to treat should be reduced to the possible minimum in
order to minimize the costs of waste treatment. To minimize the amount of aqueous
phase, the amount and quality of phosphoric acid fed to the mixer need to be optimized
by performing laboratory experiments. The refinery has a wastewater treatment plant
and the applicability of already existing treatment processes should be studied. If the
aqueous phase cannot be treated utilizing own processes, purchasing the treatment
from a specialized waste treatment company can be the best option.
Although the demetallization process produces waste in the form of aqueous waste
stream, the catalyst waste could be reduced as the catalysts are changed less frequently.
96
If the demetallization process efficiently reduced the amount of metal impurities in the
VGO, the amount of catalyst waste and the cost of treating it would significantly
decrease. Moreover, the costs of purchasing new fresh catalyst would also decrease.
9 Discussion
The HDS catalyst deactivates fast in both cases 1 and 2. Due to the rapid deactivation, a
pretreatment process is required if a VGO feed with 10 ppm of metal impurities is used.
However, the demetallization process utilizing phosphoric acid does not remove
sufficiently metal impurities based on the experimental data available. More
experimental data would be required to optimize the process conditions and to
determine the actual removal rates of metal impurities in the demetallization process.
The average cost per year caused by the shutdowns is presented in Table 27. In both
cases, the reactors are packed with catalysts before the operation is started. The lost
profit forms the largest share of the shutdown costs.
Table 27. Average annual cost due to the catalyst change shutdowns.
Process Number of shutdowns
during one year Total annual cost of
changing the catalyst (M€)
Case 1: without demetallization 5.5 31.5
Case 2: with demetallization 5 28.7
The costs caused by the shutdowns are enormous making the investment cost of the
demetallization process a substantially small part of the total costs. Adding the
demetallization process increases both the amount of equipment and the maintenance
costs related to equipment. However, as the static mixer and decanter do not include
97
moving parts, the maintenance costs of demetallization equipment are relatively low
compared to the maintenance costs of other equipment in the HDS process.
According to the sensitivity analysis in chapter 8.3.3, the number of shutdowns must
decrease or the difference in purchase costs between low-quality and cleaner VGO must
increase in order to make the demetallization process profitable. However, achieving
sufficiently high removal rates with phosphoric acid to make the process feasible is
improbable based on the available data. Moreover, the degree to which the phases can
be separated needs to be studied. As the amount of acid is minuscule compared to the
amount of VGO, separating the dispersed phase in a decanter may be problematic. If the
phases are not completely separated, more impurities enter the HDS reactors shortening
the catalyst lifetime.
10 Conclusions
The aim of this thesis was to study possible impurity removal methods for VGO prior to
the HDS process. An impurity removal pretreatment could enable the use of cheaper,
poor-quality VGO which has high impurity levels.
In the literature part of this thesis, impurities in VGO and different impurity removal
methods were discussed. Based on the advantages and disadvantages of the methods,
demetallization process utilizing phosphoric acid was chosen to be preliminarily designed
in the applied part.
VGO feed with 10 ppm of metal impurities was used in the HDS process. Without the
demetallization process, the HDS catalyst lifetime was only 8 weeks. Due to the rapid
deactivation, a pretreatment process is required if a VGO feed with 10 ppm of metal
impurities is used. Integrating the demetallization process into the HDS extended the
catalyst lifetime by one week resulting in a catalyst lifetime of 9 weeks. The catalyst
98
changing shutdowns would occur 5 times each year. The demetallization process utilizing
phosphoric acid does not remove sufficiently metal impurities based on the
experimental data available.
The cost of one process shutdown was 5733 k€. The costs caused by the shutdowns were
enormous mainly due to lost profit. The investment cost of the demetallization process
was 2298 k€ which formed a relatively small part of the total costs as the shutdowns
occurred frequently. The investment cost did not affect the profitability of the process
significantly. However, by reducing the number of shutdowns by 17 % or increasing the
purchase cost difference between low-quality and cleaner VGO by 20 %, the net present
value of the demetallization became positive. Based on the available data, achieving
sufficiently high removal rates with phosphoric acid to make the process feasible is
improbable.
More experimental data is required to determine the optimal operating conditions and
the actual removal rates. Also more experiments are needed to study if the phase
separation is possible as the amount of H3PO4 is extremely small compared to the
amount of VGO.
99
Bibliography
Abbas, S., Maqsood, Z. & Ali, M., The Demetallization of Residual Fuel Oil and Petroleum
Residue, Petroleum Science and Technology 28 (2010) 1770-1777.
Adams, C., Ellert, H., Kimberlin Jr., C. & Hamner, G., Demetallization with hydrofluoric
acid, Patent No. US3203892 A, 1965.
Admix, Sizing the Admixer Static Mixer and Sanitary Static Blender,
http://www.admix.com/pdfs/admixer-tech102.pdf, Accessed 14 March 2016.
Airgas, Safety Data Sheet, Hydrogen Fluoride,
https://www.airgas.com/msds/001077.pdf, Accessed 13 November 2015.
Ali, M. & Abbas, S., A review of methods for the demetallization of residual fuel oils, Fuel
Processing Technology 87 (2006) 573-584.
American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., ASHRAE
Handbook - Heating, Ventilating, and Air-Conditioning Systems and Equipment, American
Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., 2012, p. 31.14.
Arnold, K. & Stewart, M., Surface Production Operations - Design of Oil Handling Systems
and Facilities, Volume 1, Elsevier, 2008, pp. 377-383.
Bahadori, A., Essentials of Coating, Painting, and Lining for the Oil, Gas, and
Petrochemical Industries, Elsevier, 2015, pp. 297-300.
Bertoncini, F., Courtiade-Tholance, M. & Thiébaut, D., Gas Chromatography and 2D-Gas
Chroma for Petroleum Industry - The Race for Selectivity, Editions Technip, 2013, pp. 2-3.
Bonné, R., van Steenderen, P. & Moulijn, J., Hydrogenation of nickel and vanadyl
tetraphenylporphyrin in absence of a catalyst: A kinetic study. Applied Catalysis A:
General 206 (2001) 171-181.
100
Bröll, D., Kaul, C., Krämer, A., Krammer, P., Richter, T., Jung, M., Vogel, H. & Zehner, P.,
Chemistry in Supercritical Water. Angewandte Chemie (International ed.) 38 (1999)
2998-3014.
Chainet, F., Courtiade, M., Lienemann, C.-P., Ponthus, J. & Donard, O., Silicon speciation
by gas chromatography coupled to mass spectrometry in gasolines. Journal of
Chromatography A 1218 (2011) 9269-9278.
Chainet, F., Le Meur, L., Lienemann, C.-P., Ponthus, J., Courtiade, M. & Donard, O.,
Characterization of silicon species issued from PDMS degradation under thermal
cracking of hydrocarbons: Part 1 – Gas samples analysis by gas chromatography-time of
flight mass spectrometry. Fuel 111 (2013) 519-527.
Chemical Engineering, Plant Cost Index, http://www.chemengonline.com/pci, Accessed
17 February 2016.
Coker, A., Modeling of Chemical Kinetics and Reactor Design, Elsevier, 2001, pp. 597-600.
Coker, A., Ludwig's Applied Process Design for Chemical and Petrochemical Plants,
Volume 1, Elsevier, 2007, pp. xxv, 506-514.
Couper, J., Penney, W., Fair, J. & Walas, S., Chemical Process Equipment - Selection and
Design, 3rd edition, Elsevier, 2012, p. 305.
Davis, J., Corrosion - Understanding the Basics, ASM International, 2000, pp. 220-225.
Dechaine, G. & Gray, M., Chemistry and Association of Vanadium Compounds in Heavy
Oil and Bitumen, and Implications for Their Selective Removal, Energy Fuels 24 (2010)
2795-2808.
Eidem, P., Reducing the metals content of petroleum feedstocks, Patent No. US4752382
A, 1988.
101
Escoe, A., Pressure Vessel and Stacks Field Repair Manual, Elsevier, 2008, pp. 11-14, 25-
26.
European Parliament & Council of the European Union, Directive 2009/30/EC of the
European Parliament and of the Council, 2009.
Fahim, M., Al-Sahhaf, T. & Elkilani, A., Fundamentals of Petroleum Refining, Butterworth-
Heinemann, 2010, pp. 19-20, 76-82, 155, 167-169, 327-328.
Fine Tubes, Alloy 904L (UNS N08904).
http://www.finetubes.co.uk/uploads/docs/Alloy_904L.pdf, Accessed 18 March 2016.
Furimsky, E. & Massoth, F., Deactivation of hydroprocessing catalysts, Catalysis Today 52
(1999) 381-495.
Furimsky, E. & Massoth, F., Hydrodenitrogenation of Petroleum, Catalysis Reviews 47
(2005) 297-489.
Gaile, A., Semenov, L., Varshavskii, M., Erzhenkov, A., Koldobskaya, L. & Kaifadzhyan, E.,
Extraction Refining of Light Vacuum Gas Oil, Russian Journal of Applied Chemistry 74
(2001) 325-329.
Garverick, L., Corrosion in the Petrochemical Industry, ASM International, 1994, pp. 321-
329.
Gawel, I., Bociarska, D. & Biskupski, P., Effect of asphaltenes on hydroprocessing of heavy
oils and residua, Applied Catalysis A: General 295 (2005) 89-94.
Gerrard, M., Guide to Capital Cost Estimating, IChemE, 2000, p. 85.
Gorra, F., Scribano, G., Christensen, P., Vibeke Andersen, K. & Gaetano Corsaro, O., New
catalyst, improved presulfiding result in 4+ year hydrotreater run, Oil & Gas Journal 91
(1993) 39-43.
102
Hall, S., Rules of Thumb for Chemical Engineers, Elsevier, 2012, pp. 150-151, 163.
Hurme, M., Process Design Manual, 2008, 107 p.
Jechura, J., Refinery Feedstocks & Products - Properties & Specifications.
http://inside.mines.edu/~jjechura/Refining/02_Feedstocks_&_Products.pdf, Accessed 4
February 2016.
Jones, D. & Pujadó, P., Handbook of Petroleum Processing, Springer, 2006, pp. 169-171.
Juyal, P., Mapolelo, M., Yen, A., Rodgers, R. & Allenson, S., Identification of Calcium
Naphthenate Deposition in South American Oil Fields, Energy Fuels 29 (2015) 2342-2350.
Kauppalehti, Valuutta: USA dollari,
http://www.kauppalehti.fi/5/i/porssi/valuutat/valuutta.jsp?curid=USD, Accessed 17
February 2016.
Kimberlin Jr., C. & Judson, M., Removal of metal contaminants from catalytic cracking
feed stocks with sulfuric acid, Patent No. US2902430 A, 1959.
Krambeck, F., Lam, C. & Schipper, P., Process for removing metals from crude, Patent No.
US4645589 A, 1987.
Kritzer, P., Corrosion in high-temperature and supercritical water and aqueous solutions:
a review, The Journal of Supercritical Fluids 29 (2004) 1-29.
Kuehne, D., Hawker, L. & Kramer, D., Method for removing calcium from crude oil, Patent
No. US6905593 B2, 2005.
Kukes, S., Demetallization of hydrocarbon containing feed streams with phosphorous
compounds, Patent No. US4529503 A, 1985.
Kukes, S. & Battiste, D., Demetallization of heavy oils with phosphorous acid, Patent No.
US4522702 A, 1985.
103
Laredo, G., De los Reyes, J., Cano, J.-L. & Castillo, J., Inhibition effects of nitrogen
compounds on the hydrodesulfurization of dibenzothiophene, Applied Catalysis A:
General 207 (2001) 103-112.
Lee, J., Shin, S., Ahn, S., Chun, J., Lee, K., Mun, S., Jeon, S., Na, J. & Nho, N., Separation of
solvent and deasphalted oil for solvent deasphalting process, Fuel Processing Technology
119 (2014) 204-210.
Leprince, P., Petroleum Refining, Volume 3 - Conversion Processes, Editions Technip,
2001, pp. 416-417, 534-536.
Macaud, M., Sévignon, M., Favre-Réguillon, A., Lemaire, M., Schulz, E. & Vrinat, M., Novel
Methodology toward Deep Desulfurization of Diesel Feed Based on the Selective
Elimination of Nitrogen Compounds, Industrial and Engineering Chemistry Research 43
(2004) 7843-7849.
Management Engineering & Production Services, World carbon steel prices,
http://www.meps.co.uk/World%20Carbon%20Price.htm, Accessed 18 February 2016.
Mandal, P., Diono, W., Sasaki, M. & Goto, M., Nickel removal from nickel etioporphyrin
(Ni-EP) using supercritical water in the absence of catalyst, Fuel Processing Technology
104 (2012) 67-72.
Mandal, P., Goto, M. & Sasaki, M., Removal of Nickel and Vanadium from Heavy Oils
Using Supercritical Water, Journal of the Japan Petroleum Institute 57 (2014) 18-28.
Manning, F. & Thompson, R., Oilfield Processing Volume Two: Crude Oil, PennWell
Publishing Company, 1995, p. 47.
Marcilly, C., Acido-Basic Catalysis, Volume 1 - Application to Refining and Petrochemistry,
Editions Technip, 2006, pp. 326-327.
McKetta, J., Petroleum Processing Handbook, Marcel Dekker, Inc., 1992, pp. 533-539.
104
Menon, E., Transmission Pipeline Calculations and Simulations Manual, Elsevier, 2015,
pp. 156-158.
Moss, D. & Basic, M., Pressure Vessel Design Manual, Elsevier, 2013, pp. 186-188.
NFPA, Comparison of NFPA 704 and HazCom 2012 Labels,
http://www.nfpa.org/Assets/files/AboutTheCodes/704/NFPA704_HC2012_QCard.pdf,
Accessed 13 November 2015.
Northeastern University, NFPA HAZARD RATING SYSTEM,
http://www.ehs.neu.edu/laboratory_safety/general_information/nfpa_hazard_rating/,
Accessed 13 November 2015.
Panossian, Z., Lira de Almeida, N., Ferreira de Sousa, R., de Souza Pimenta, G. & Bordalo
Schmidt Marques, L., Corrosion of carbon steel pipes and tanks by concentrated sulfuric
acid: A review, Corrosion Science 58 (2012) 1-11.
Parisher, R. & Rhea, R., Pipe Drafting and Design, Elsevier, 2012, pp. 121-123.
Parkash, S., Refining Processes Handbook, Gulf Professional Publishing, 2003, pp. 16-24,
197-203.
Pauling, L., General Chemistry, Dover Publications, 1970, pp. 279-282.
Pohanish, R., Sittig's Handbook of Toxic and Hazardous Chemicals and Carcinogens,
Elsevier, 2012, pp. 2147-2153.
PotashCorp, Q4 Market Analysis Report,
www.potashcorp.com/media/MAR_Q4_2014.pdf, Accessed 26 February 2016.
Powell, R., Removal of metal components from petroleum oils, Patent No. US2778777 A,
1957.
105
Rand, S., Significance of Tests for Petroleum Products, ASTM International, 2010, pp. 48-
49.
Riazi, M., Characterization and Properties of Petroleum Fractions, ASTM International,
2005, pp. 129-130, 141-142.
Ropital, F., Corrosion and Degradation of Metallic Materials - Understanding of the
Phenomena and Applications in Petroleum and Process Industries, Editions Technip,
2009, pp. 27-33, 52-55.
Sau, M., Basak, K., Manna, U., Santra, M. & Verma, R., Effects of organic nitrogen
compounds on hydrotreating and hydrocracking reactions, Catalysis Today 109 (2005)
112-119.
Schobert, H., Chemistry of Fossil Fuels and Biofuels, Cambridge University Press, 2013,
pp. 267-272.
Seider, W., Seader, J. & Lewin, D., Product and Process Design Principles: Synthesis,
Analysis and Design, John Wiley and Sons, Inc., 2004, pp. 527-532.
Shekhawat, D., Spivey, J. & Berry, D., Fuel Cells: Technologies for Fuel Processing, Elsevier,
2011, pp. 327-329.
Shiraishi, Y., Hirai, T. & Komasawa, I., Photochemical Denitrogenation Processes for Light
Oils Effected by a Combination of UV Irradiation and Liquid-Liquid Extraction, Industrial
& Engineering Chemistry Research 39 (2000) 2826-2836.
Shiraishi, Y., Hirai, T. & Komasawa, I., Photochemical Desulfurization and
Denitrogenation Process for Vacuum Gas Oil Using an Organic Two-Phase Extraction
System, Industrial & Engineering Chemistry Research 40 (2001) 293-303.
Sinnott, R., Coulson and Richardson's Chemical Engineering Volume 6 - Chemical
Engineering Design, Elsevier, 2005, pp. 815-819.
106
Speight, J., The Chemistry and Technology of Petroleum, Marcel Dekker, Inc., 1999, pp.
235-241.
Srivastava, V., An evaluation of desulfurization technologies for sulfur removal from
liquid fuels, RSC Advances 2 (2012) 759-783.
Steel Tank Institute/Steel Plate Fabricators Association, Relative Cost Ratio,
http://www.steeltank.com/Portals/0/Pressure%20Vessels/SSWseminarOct2012/Relati
Re%20Cost%204%2015%202012.pdf, Accessed 18 March 2016.
Sun, Y., Yang, C., Zhao, H., Shan, H. & Shen, B., Influence of Asphaltene on the Residue
Hydrotreating Reaction, Energy Fuels 24 (2010) 5008-5011.
Thakore, S. & Bhatt, B., Introduction to Process Engineering and Design, Tata McGraw-
Hill Publishing Company Limited, 2007, pp. 344-352.
The American Society of Mechanical Engineers, ASME Boiler & Pressure Vessel Code
Section VIII - Rules for Construction of Pressure Vessels, ASME, 2010, pp. 19-20, 29.
The American Society of Mechanical Engineers, Pipeline Transportation Systems for
Liquids and Slurries, ASME, 2012, pp. 18-21.
The American Society of Mechanical Engineers, ASME Boiler & Pressure Vessel Code
Section II – Materials, ASME, 2015.
Toulhoat, H. & Raybaud, P., Catalysis by Transition Metal Sulphides - From Molecular
Theory to Industrial Application, Editions Technip, 2013, pp. 303-306, 320-321, 618, 680,
703-704.
Towler, G. & Sinnott, R., Chemical Engineering Design - Principles, Practice and Economics
of Plant and Process Design, 2nd edition, Elsevier, 2013, 1265 p.
107
Trading Economics, United States Inflation Rate,
http://www.tradingeconomics.com/united-states/inflation-cpi, Accessed 22 March
2016.
Walas, S., Chemical Process Equipment - Selection and Design, Elsevier, 1990, pp. 663-
669.
Yaws, C., Yaws' Handbook of Properties of the Chemical Elements, Knovel, 2011.
ZAIN Technologies, Market Pricing: Static Mixers,
http://www.zmixtech.com/mixequip/static/Purchase_Static_Mixers.html, Accessed 18
March 2016.
1
Graph for determining the friction factor in pipes APPENDIX 1
Figure 1. Pipe friction factor in relation to Reynolds number and pipe roughness
(Towler & Sinnott, 2013).